Volume 20 Issue 1 by IFAMA

International Food and Agribusiness Management Review

Official Journal of the International Food and Agribusiness Management Association

Volume 20 Issue 1 2017

EVERY GENERATION NEEDS ITS LEADERS. As a farmer-owned cooperative, we help lead the agricultural industry by promoting the most innovative, responsible farming practices available today. And we want to make sure that the next generation of leaders is well trained for tomorrow. That’s why we proudly support youth leadership programs like FFA. It’s also why we’ve invested in scholarships and countless land-grant university initiatives across the country. At CHS, we know every generation needs its leaders. And every potential leader needs a little help becoming one. Visit chsinc.com/stewardship to learn more about our ongoing commitment to leadership.

International Food and Agribusiness Management Review

Editorial Staff Executive Editor

Gerhard Schiefer, University of Bonn, Germany

Regional Managing Editors Asia, Australia, and New Zealand

Derek Baker, UNE, Australia Kim Bryceson, University of Queensland, Australia Kevin Chen, IFPRI-Bejing, China Jeff Jia,University of Exeter, United Kingdom Nicola M. Shadbolt, Massey University, New Zealand

Europe

Pegah Amani, Technical Institute of Sweden, Sweden Vera Bitsch, Technical University of Munich, Germany Alessio Cavicchi, University of Macerata, Italy Hans De Steur, Ghent University, Belgium Klaus Frohberg, University of Bonn, Germany Cristina Santini, University San Raffaele, Italy Jacques Trienekens, Wageningen University, The Netherlands

North America

Ram Acharya, New Mexico State University, USA Yuliya Bolotova, Clemson University, USA Michael Gunderson, Purdue University, USA Mark Hansen, Brigham Young University, USA William Nganje, North Dakota State, USA R. Brent Ross, Michigan State University, USA Aleksan Shanoyan, Kansas State University, USA David Van Fleet, Arizona State University, USA Nicole Olynk Widmar, Purdue University, USA Cheryl Wachenheim, North Dakota State University, USA

South America

Aziz da Silva Júnior, Federal University of Vicosa, Brazil Jose Vincente Caixeta Filho, University of Sao Paulo, Brazil Filippo Arfini, Universita’ di Parma, Italy Stefano Boccaletti, Universita’ Cattolica, Italy Michael Boehlje, Purdue University, USA Dennis Conley, University of Nebraska - Lincoln, USA Francis Declerck, ESSEC Business School, France Jay Lillywhite, New Mexico State University, USA Woody Maijers, INHOLLAND University, The Netherlands

Marcos Fava Neves, FEA / USP / PENSA, Brazil Onno Omta, Wageningen University, The Netherlands Hernán Palau, Buenos Aires University, Argentina Christopher Peterson, Michigan State University, USA Thomas Reardon, Michigan State University, USA Mary Shelman, (Retired) Harvard Business School, USA Johan van Rooyen, University of Stellenbosch, South Africa

The IFAMR (ISSN #: 1559-2448) is published quarterly and the archived library is available at http://www.ifama.org. For copyright and publishing information, please contact: Marijn van der Gaag, Administrative Editor Wageningen Academic Publishers • P.O. Box 220 6700 AE Wageningen • The Netherlands • Tel: +31 317 476511 Fax: +31 317 453417 • Email: ifamr@wageningenacademic.com • Web: http://www.wageningenacademic.com/loi/ifamr

International Food and Agribusiness Management Review Volume 20 Issue 1, 2017

TABLE OF CONTENTS 1.

The responsibility of global agribusiness: consequences for agribusiness research

Business models for maximising the diffusion of technological innovations for climate-smart agriculture

Assessment of socio-economic configuration of value chains: a proposed analysis framework to facilitate integration of small rural producers with global agribusiness

The supply chain of Brazilian maize and soybeans: the effects of segregation on logistics and competitiveness

Predicting grower choices in a regulated environment

An application of activity-based costing in the chicken processing industry: a case of joint products

Food safety as a field in supply chain management studies: a systematic literature review

Gerhard Schiefer

Thomas B. Long, Vincent Blok, and Kim Poldner

Miguel Arato, Stijn Speelman, Joost Dessein, and Guido van Huylenbroeck

Andréa L.R. de Oliveira and Augusto M. Alvim

5. 6.

Juan S. Castillo-Valero, Mercedes Sánchez-García, and Mari Carmen García-Cortijo

Panravee Kabinlapat and Siriluck Sutthachai

Daniel P. Auler, Rafael Teixeira, and Vinícius Nardi

8. 9.

Food scare crisis: the effect on Serbian dairy market

Rade Popovic, Boris Radovanov, and James W. Dunn

113

Food safety and food imports in Europe: the risk of aflatoxins in pistachios 129

Bo Xiong

International Food and Agribusiness Management Review

Table of contents

Volume 20, Issue 1, 2017

10. Territory, environment, and healthiness in traditional food choices: insights into consumer heterogeneity

143

11. Investigating the impact of maximum residue limit standards on the vegetable trade in Japan

159

Fabio Boncinelli, Caterina Contini, Caterina Romano, Gabriele Scozzafava, and Leonardo Casini

Jong Woo Choi and Chengyan Yue

International Food and Agribusiness Management Review

OPEN ACCESS International Food and Agribusiness Management Review Volume 20 Issue 1, 2017; DOI: 10.22434/IFAMR2017.x001

The responsibility of global agribusiness: consequences for agribusiness research EDITORIAL

The agri-food sector, like other industries, is currently confronted with several broad issues that impose either new constraints or new goals on the sector. Foremost among these issues are certainly various environmental concerns, such as climate change, increasing water scarcity in some regions of the world, or the loss of arable land. New goals for agribusiness stem, for example, from the gentrification of the populations of many rich countries, the rapidly growing urbanization of all countries, and the spreading obesity in many societies. These issues are undeniably important and deserve close attention by the sector. The issues should, however, not lead us to forget that the core task of agribusiness is to assure food security for all people in the world, including the two additional billions that will populate the planet by 2050 or so. The prospect of continued world population growth invokes in the minds of many the dark spirits of Malthus. But Malthus’ curse has lost its spell. Given our state of knowledge, assuring food security seems easy. We are assured by one of the world’s leading students of famine, ‘The prospect of a famine-free world hinges on improved governance and peace. It is as simple – or difficult – as that’ (Ó Gráda, 2009: 2821). As many economists would tell us, when food markets are competitive and the sector is well-governed, the commercial spirits of agribusiness alone will assure that food supplies will satisfy consumers’ demand. But are world food markets competitive, and are they well governed? Neither is assured. Good governance requires that there is (a) some feedback about system performance in terms of the quantity, quality, diversity and affordability for consumers of food supplies, and (b) the ability of the system to react without undue delay to inadequate performance. Both the flow of information and the activities triggered by the information are embedded in an organization which is best described as an internet of supply networks that extend beyond national borders. The ability of the networks to gather, condense, and disseminate information about the agri-food system’s performance is not in doubt. There are innumerable interconnected eyes and ears that perceive food system performance failures whenever and wherever they may occur, and food commodity exchanges register food shortages in advance, mostly with high accuracy and reliability. Will the sector act responsibly to the available information? Agribusiness agents, the ultimate industry actors, are engaged in the delivery of food to consumers in a variety of specialized activities reaching from production agriculture to processing, transportation, and distribution. Individually, the specialists cannot be obliged to assure billions of consumers that their supply of food, in terms of quantity, quality, diversity and affordability is secure. In contrast to individual agri-food industry agents, the sector as a whole cannot be relieved from its obligation to assure food security for all. Its performance is an endogenous feature of the 1Ó

Gráda, C. 2009. Famine: a short history. Princeton University Press, Princeton, NJ, USA.

Editorial

Volume 20, Issue 1, 2017

sector and the question is whether the supply networks that constitute the sector can be relied upon to meet that obligation. The networks are stable patterns in the commercial transactions of agribusiness agents. The governance of networks is, however, mostly distributed and networks with uncontested captains are rare. And most importantly, there are in the networks no identifiable single agents or groups of agents who take responsibility if their network fails to meet its food supply obligations. In short, at the network level there is no responsibility to match the obligation. Without this responsibility the networks may or may not respond adequately to actual or expected deficiencies in assuring food security for consumers on global scale. Unfortunately, we do not know the circumstances under which a food supply network will respond adequately to a lack of performance and when such response may not be forthcoming. There is another feature of networks that gives rise to concern. Networks are stable but not permanent arrangements, they may disintegrate. The dissolution of a food supply network is of no great concern when another quickly absorbs all or most of the agents that have been orphaned by the dissolved one. As food supply networks grow under the joint impact of economies of scale, of network economies from standardization, and of information economies, there may be few or no networks to adopt agents that have been orphaned by a large network. We know very little about the conditions under which large networks unravel. Some simulation studies suggest that removal of agents, that is network shrinkage, may induce network instability and eventual dissolution. For us as agribusiness researchers the two threats to agribusiness for meeting its food supply obligation responsibly foremost imply two things. First, we need to engage in research on governance systems that allow food supply networks to adapt to actual and expected threats to their ability to meet their food supply obligations to society. Second, we need to understand the conditions under which networks tend to dissolve. Arguably and given the global challenge of food security, such research should have at least equal priority over research on the efficiency, or sustainability of agribusiness. Moving the sectorâ&#x20AC;&#x2122;s activities beyond sustainability towards meeting its obligation may not be feasible without compromising sustainability concerns. For overcoming this barrier the sector may need to intensify its cooperation with policy makers, civil society groups, business associations, marketing groups, business representatives, as well as between individual enterprises. New kinds of responsibility networks may be needed for this task and agribusiness research may have a role in their design and evolution. The International Food and Agribusiness Management Association and other associations in the field could provide platforms for initiating such movements. As global meeting points for research, industry and policy they provide the necessary actor mix and the necessary competence for understanding future scenarios and for moving forward with initiatives. This reaches beyond research meetings and discussions and requires intensive links between research and the various actors in the field. The responsibility of the sector for assuring food security translates into a responsibility of associations for supporting the sector in this endeavor and for providing a strong â&#x20AC;&#x2DC;platform infrastructureâ&#x20AC;&#x2122; that allows sector agents of all kinds to leverage their combined competence and to moving forward decisively. It is our understanding that there is much room for improvements in living up to this responsibility. As one very first step in this direction, two international journals, the International Food and Agribusiness Management Review (IFAMR) and the Journal on Chain and Network Science have joined ranks in a restructured IFAMR which is supported by a new publisher (Wageningen Academic Publishers) with high visibility in the agri-food community. With this move, the journal is prepared to cover the broad range

International Food and Agribusiness Management Review

Editorial

Volume 20, Issue 1, 2017

of issues the sector has to deal with including the challenges faced by the sector in the organization and management of the diversity of existing and emerging chains and networks on which the sector depends. The journal encourages the scientific community to utilize its competence and to contribute to the discussion through the analysis of problems and the clarification of implications for the sector and its actors. This might ask to not only look at the past through statistical analysis but to engage in forward looking scenariooriented research. The journal is open for developing new formats that might help to communicate research and management issues that need further discussion in the communities. Our profession is rich in theories, models, experiences and thoughts which should find its place in the journal for reaching out to all those concerned in research and sector development who are needed for supporting the sector in realizing its goals and responsibilities. Gerhard Schiefer, Executive Editor2

I thank Rolf A.E. Mueller of the University at Kiel for helpful discussions.

International Food and Agribusiness Management Review

OPEN ACCESS International Food and Agribusiness Management Review Volume 20 Issue 1, 2017; DOI: 10.22434/IFAMR2016.0081 Received: 6 April 2016 / Accepted: 12 September 2016

Business models for maximising the diffusion of technological innovations for climate-smart agriculture RESEARCH ARTICLE Thomas B. Long a, Vincent Blokb, and Kim Poldnerc aPostdoctoral

fellow in Business Models for Sustainability, Management Studies, Wageningen University, Hollandseweg 1, 6706 KN, Wageningen, the Netherlands bAssociate

Professor in Sustainable Entrepreneurship, Business Ethics and Responsible Innovation, Management Studies, Wageningen University, Hollandseweg 1, 6706 KN, Wageningen, the Netherlands cAssistant

Professor, Management Studies, Wageningen University, Hollandseweg 1, 6706 KN, Wageningen, the Netherlands

Abstract Technological innovations will play a prominent role in the transition to climate-smart agriculture (CSA). However, CSA technological innovation diffusion is subject to socio-economic barriers. The success of innovations is partly dependent on the business models that are used to diffuse them. Within the context of innovations for CSA, the role that innovation providers’ business models play in the successful adoption and diffusion has received limited attention. In this paper we identify critical issues for business models for CSA technological innovations (BMfCSATI). Our results indicate that current BMfCSATIs are not optimised for diffusing CSA technological innovations. Critical business model elements include the value proposition, channels, customer relationships, key resources, key partners, and cost structure. We find a disparity between the views of CSA technological innovation providers and potential users. The paper explores the implications of the results and develops recommendations for CSA technological innovation providers’ business models. Keywords: business models for sustainable innovation, business models for sustainability, business model canvas, Europe, new technology based firms, technological innovations JEL code: Q16, O00, O32 Corresponding author: thomas.long@wur.nl

Long et al.

Volume 20, Issue 1, 2017

1. Introduction The development, adoption and diffusion of pro-environmental technological innovations is critical for enhancing sustainability (EIT, 2014; European Commission, 2014a; Montalvo, 2008). This is also the case with agriculture. Agriculture will adapt to changes in weather patterns associated with climate change, and plays a role in limiting greenhouse gas (GHG) emissions (Coumou et al., 2014; Kurukulasuriya and Rosenthal, 2003; Trnka et al., 2014) whilst feeding future global population (Bogdanski, 2012; Nelson, et al., 2009). These challenges mean that the development, adoption and diffusion of appropriate technological innovations is an urgent priority. However, the adoption of technological innovations, including within agricultural contexts, can often be impacted by a range of socio-economic barriers. These barriers impact both the production and marketing of technological innovations, as well as their adoption and use (Montalvo, 2008). Climate-smart Agriculture (CSA) is a programmatic response to these challenges, seeking to encourage sustainable increases in agricultural productivity and incomes, the building of resilience and adaptation to climate impacts as well as reduction of GHG emissions where possible (FAO, 2010, 2014). As such, innovations consistent with these principles need to be adopted by agri-food chains. The successful adoption and diffusion of technological innovation is dependent upon many factors. The support of appropriate and effective business models is noted as one promising strategy for enhancing the success of technological innovations (Boons and LĂźdeke-Freund, 2013; Teece, 2010). Business model innovation has also been identified as a critical component of the transition to a sustainable future (Hansen et al., 2009). By examining the business models that support technological innovation, key organisational factors that promote or inhibit adoption and diffusion can be identified and explored. In turn, this allows interventions, in terms of changes to business strategy or policy, to be designed. This is especially pertinent for technological innovations for agriculture, as previous business model approaches within this context have to date only considered the user or adopter perspective (Sivertsson and Tell, 2015), or only within developing country contexts (Chesbrough et al., 2006). In this paper, we identify critical issues for the business models of CSA technological innovation providers. We do this by exploring the barriers of both CSA technological innovation providers and potential users and what they think could enhance diffusion. These factors are then applied to a business model framework, the business model canvas (BMC). Through this process, we show that the current business models employed by CSA technological innovation providers are not optimised to current market demands, and as such, can be seen to be inhibiting the adoption and diffusion of CSA technological innovations. We are guided by the question: what are the critical issues of CSA technological innovation business model development for the adoption and diffusion of CSA technological innovations? In order to answer this question, we review previous research on business models and their relationship to innovation and sustainable innovation. Generic critical issues are drawn out from the literature on business models for sustainable innovation (BMfSI); these are mapped onto the much used BMC (Osterwalder and Pigneur, 2009) to create a theoretical framework. This research is carried out within the context of the CSA Booster, a Climate-KIC funded European project investigating the adoption and diffusion of CSA technological innovations. European technological innovations providers with innovations consistent with the principles of CSA (i.e. that they enhance agricultural productivity and either contribute to climate adaption or mitigation), formed the focus of the empirical investigation. Climate change is high on the agricultural policy agenda in the European Union, with actions including European Commission strategies to encourage member states to â&#x20AC;&#x2DC;climate proofâ&#x20AC;&#x2122; their agricultural sectors and improve decision-making (European Commission, 2014b).

International Food and Agribusiness Management Review

Long et al.

Volume 20, Issue 1, 2017

2. Literature review Business models Business models emerged as a concept in the late 1990s (Shafer et al., 2005). They describe how organisations create value, select customers, assign processes and enter markets, and can be used for analysis, comparison, management and innovation (Benijts, 2014; Doganova and Eyquem-Renault, 2009; Osterwalder et al., 2005). Many business model frameworks have been developed to aid the analysis and description of business models, with examples including the component business model (Pohle et al., 2006), business reference model (Fettke and Loos, 2007) and the much used and publicised BMC (Osterwalder and Pigneur, 2009). The core elements of a business model (Boons and Lüdeke-Freund, 2013; Chesbrough and Rosenbloom, 2002; Osterwalder et al., 2005; Osterwalder and Pigneur, 2009) include the identification or articulation of the: ■■ value proposition, involving how value is generated; ■■ customer segment, highlighting who the users or customers will be; ■■ customer relationships, focusing on how the business engages with its customers; ■■ channels, which highlights how customers are reached, including awareness raising and the provision of information; ■■ key activities, articulating activities required to carry out the other business model functions; ■■ key resources, concerning the critical assets needed; ■■ key partners, highlighting those actors that are critical to delivery of the value proposition; ■■ cost structure and the revenue streams, outlining key costs and how an organisation generates revenues. Business model frameworks can be used to explore and plan how an organisation will operate and compete, or to identify areas that need improvement; this last aspect is especially pertinent to the case in hand. Research on business models can be categorised into three distinct streams (Wirtz, 2011). First, in conjunction with the ‘dotcom’ boom around the turn of this century, research focuses on the interaction between technology and business models. For example, how the internet changed revenue models, or the relative success of innovations due to business model choices (Chesbrough and Rosenbloom, 2002). A second stream looks at how businesses organise or structure themselves. The third stream is more strategically focused, considering how organisations compete through the evolution of their value proposition and business model innovation (Chesbrough, 2010). Business model innovation is a common and often dominant topic across these three strands of research (Boons and Lüdeke-Freund, 2013). On the one hand, technological innovation is explored in terms of how new products and services interact with existing business models. This contrasts with investigations of how the innovation of business models themselves can extract extra value from existing products or services (Baden-Fuller and Morgan, 2010; Wirtz, 2011). This presents three combinations for business models and innovation: either an innovative business model for existing products, an existing business model for an innovation or an innovative business model for innovative products. Conventional economic and management theory assumes that a good innovation will succeed in the market. This assumption however is increasingly challenged by business model approaches. Many business model scholars assert that innovations often require innovative business models in order to be diffused and be adopted successfully (Bohnsack et al., 2014; Boons et al., 2013; Teece, 2010). This may be the case with regards to CSA technological innovations and serves as a strong rationale for examining critical issues for business models for CSA technological innovations (BMfCSATIs). In summary, the specific business models used by technology providers impacts the relative success of these innovations in the market. Business models can act as a common language within innovation networks, allowing actors to identify and discuss opportunities for the commercialisation of innovation (Boons and Lüdeke-Freund, 2013). Business models can become ‘market devices’ (Doganova and Eyquem-Renault, 2009), able to boost the diffusion of International Food and Agribusiness Management Review

Long et al.

Volume 20, Issue 1, 2017

emerging innovations by overcoming barriers (Wells, 2008), and connecting production with consumption (Boons and LĂźdeke-Freund, 2013; Iles and Martin, 2013; Wells, 2008). This is an important point, as CSA technological innovations experience socio-economic barriers to their diffusion. A business model perspective may identify deficiencies in the current approaches of CSA technological innovation providers and highlight ways to boost CSA technological innovation diffusion. Indeed, business model innovation is seen to be able to counter several barriers associated with the development and diffusion of innovations generally (Chesbrough, 2010). These barriers can include issues such as low initial margins, making competition with established technologies more difficult. In addition, the innovation may target new and different customers and distribution channels, meaning greater uncertainty when compared with established technologies, requiring actors to challenge dominant logics and norms. In order to become successful, technology providers will have to develop compelling value propositions, advantageous cost structures, the ability to capture value and consider how the innovation will interact or be used by customers (Teece, 2010). Many of these factors have been identified as barriers in relation to CSA technological innovations. Innovation and sustainability Eco-innovations add economic value whilst simultaneously reducing environmental impacts (Horbach et al., 2012). CSA technological innovations are consistent with this definition due to the improved productivity and/or reduced GHG emissions embodied within CSA principles. Similarly, sustainable innovations are those that take into account environmental, social and economic considerations in their development and use (Larson, 2011). From this, we can conceive of sustainable innovations, eco-innovations and CSA technological innovations as having many similarities, meaning literature that examines eco-innovation, or sustainable innovations and business models, may shed light on the critical issues for BMfCSATIs as well. As shown, the success of an innovation depends, at least in part, on the development of a business model that is able to support its adoption and diffusion. Wider factors that impact the success of eco-innovations and sustainable innovations will have a corresponding impact on how business models are designed so that they are optimised for adoption and diffusion. For instance, consumer variables impact the success of sustainable innovations. Factors such as attitudes and cognitive processes affect adoption and relate to contextual and demographic factors. For example, Bhate and Lawler (1997) found that psychographic and situational variables, such as feelings towards the environment, rather than demographic factors, such as age or educational level, had the greatest impact on consumer purchasing decisions. More practically, Lin et al. (2013) note that sustainable innovations must meet user needs to be adopted and successful. Price and quality are also found to be critical factors for sustainable innovations. The price competitiveness of a sustainable innovation, compared to standard products, can be critical to its success (Brouhle and Khanna, 2012). Similarly, quality is a premium consumer concern, meaning that if a sustainable innovation does not compare favourable in this regard to other products, it is unlikely to be successful (BrĂŠcard et al., 2009). The influences of knowledge and information flows have also been found to affect the rate of diffusion of sustainable innovations (Lee et al., 2006). Specifically, the media can increase the demand for sustainable innovations by highlighting environmentally detrimental practices and effects. Stakeholder inclusion in the development of sustainable innovations also impacts success, as this can increase demand and market acceptance of new sustainable products (Byrne and Polonsky, 2001; Carrillo-Hermosilla et al., 2010). A further critical factor is a well-developed and effective delivery chain (Jabbour, 2008; Jabbour et al., 2013). These factors illustrate consumer or demand side factors, which in turn must be taken into account when designing business models that support the adoption and diffusion of sustainable innovations.

International Food and Agribusiness Management Review

Long et al.

Volume 20, Issue 1, 2017

Business models for sustainable innovation A well-defined body of research exists which focuses on the role of business models for sustainable innovation. Having considered the more general business model to innovation relationship, and explored CSA technological innovations in relation to sustainable innovation, we will now consider business models for sustainable innovation. After this the potential critical factors for CSA technological innovation providers will be synthesised on the BMC. As shown above, many factors can impact the success of sustainable innovations, but a business model perspective allows the form and operation of firms to be examined and connected to the performance of sustainable innovations. A key contribution in terms of BMfSI is provided by Boons and Lüdeke-Freund (2013), who synthesise research on sustainable innovation and sustainable business models. It is highlighted that BMfSI can be conceived at three levels. First, at the organisational level, business models illustrate how an innovation connects to other firm functions, such as marketing. This perspective is noted as somewhat neglected, with much research seeing firms as ‘black boxes’. An innovation must be marketable and so the innovation process should include firm functions such as in sales and marketing. This perspective encourages us to ask whether CSA technological innovation providers have CSA technological innovations that are marketable and attractive to potential users. Or if the products are not marketable or attractive, if this is restricting adoption and diffusion. These points draw attention to the role of an effective value proposition in boosting the diffusion of innovations. Second, the inter-organisational level of analysis draws attention to interactions within the supply chain and other external factors. This highlights business model aspects such as channels (used to link to customers), customer relations and key partners. For sustainable innovations, this highlights principles consistent with sustainable supply chain management (Seuring and Müller, 2008), such as encouraging suppliers and downstream actors, including customers, to act sustainably. The inter-organisational level also highlights how an innovation interacts with users more generally, in terms of how easy it is to adopt (Kemp and Volpi, 2008). This raises questions as to whether CSA technological innovation providers are suitably connected with potential customers and wider networks in the supply chain. The third level – the societal level – links to systems-levels thinking for sustainable transitions (Geels, 2005). The question here is the extent to which the sustainable innovation offers value to society as a whole. This level of analysis is likely to have limited value, as we focus on CSA technological innovation providers – at the actor or organisational level – in this paper. Several broad and generic normative requirements have been developed for business models aiming to provide sustainable innovations (Boons and Lüdeke-Freund, 2013). It is highlighted that the value proposition should include environmental and/or social aspects in addition to economic elements; that the supply chain includes suppliers who take on responsibility towards both their own and the focal organisation’s stakeholders, in line with sustainable supply chain management; and that the customer interface should encourage and motivate consumers to take on responsibility for their own and wider stakeholder actions. For both the supply chain and customer interface, the focal company should not just shift their responsibilities onto other actors, but rather ensure that additional and relevant responsibilities are induced. Finally, the financial model should include social and environmental externalities and ensure a fair distribution between relevant stakeholders. These normative requirements are purposefully generic and do not specify a particular business model; rather, they must be adapted to specific contexts or objectives, in this case CSA. For less generic, practical examples, several case studies on success factors for BMfSI are noteworthy. Through an examination of an eco-efficient product-service system innovation, Ceschin (2013) stresses that some financial or institutional protection can allow experimentation to take place, which could for example involve investors who are willing to take a longer-term view and forego short-term returns. Such support International Food and Agribusiness Management Review

Long et al.

Volume 20, Issue 1, 2017

may also be needed by a wider set of stakeholders. The innovation should also be developed with a clear vision, as this allows the expectations of key stakeholders to convergence, providing strategic direction. An examination of the preferences of renewable energy investors showed that the value proposition should include either ‘best service’, ‘lowest price’ or ‘best technology’ – here the value proposition is seen as the key element (Loock, 2012). Whilst a study into algae biofuel for aviation noted critical factors to include the support of a broad range of actors, a sympathetic regulatory environment, innovative and customised business models and a thorough market adoption strategy (Nair and Paulose, 2014). Critical factors for business models for climate-smart agriculture technological innovations We aim to identify critical issues for the business models of CSA technological innovations providers in this article. By reviewing and synthesising the key factors for BMfSI and wider factors impacting the success of innovations, we produce a provisional set of critical issues, mapped onto the BMC. Previous investigations into BMfSI have predominantly been normative and generic, or have focused on different contexts to agriculture. Mapping these factors onto the BMC allows us to create an initial framework (Figure 1), where we can test the applicability of wider factors empirically within the context of agriculture, and specifically CSA. In this sense, we use this business model framework as a lens through which to explore critical issues for the diffusion of technological innovations. This allows us to see which aspects of the business model relate to specific critical issues. The BMC is used extensively by practitioners, from entrepreneurs launching start-ups through to high-level decision-makers in FT Global 500 companies (Hanshaw and Osterwalder, 2015). The BMC acts firstly as a business model framework, providing a complexity-reducing backdrop for our analysis (Ching and Fauvel, 2013). We also use this framework due to its wide acceptance within the practitioner community, which we hope will make our results more widely applicable and understood.

Key partners

Key activities

Access to partners necessary to provide value proposition; i.e. suppliers, investors etc.

Value proposition

Customer relationships Customer segments

Compelling and relevant to CSA.

Ensure successful diffusion of CSATIs. Encourage wider CSA consistent behaviour.

Channels

Key resources Access to sufficient resources to provide value proposition.

Access to customers who demand CSATIs.

Cost structure

Revenue streams

Allow competitive pricing, and economic viability to the CSATI producer.

Encourage move to ‘jobs done’ rather than ‘per unit’ pricing.

Figure 1. BMfCSATIs critical issues according to the literature. BMfCSATI = business models for climatesmart agriculture technological innovation, CSA = climate-smart agriculture, CSATI = climate-smart agriculture technological innovation.

International Food and Agribusiness Management Review

Long et al.

Volume 20, Issue 1, 2017

The ‘value proposition’ articulates how value is created (Osterwalder and Pigneur, 2009). It should address the problems faced by end-users. Without a well-developed, articulated and demonstrable value proposition that meets end-user needs (Lin et al., 2013), CSA technological innovations will face difficulties in terms of marketing and sales. A poor value proposition would fail several of the ‘success factors’ identified in relation to sustainable innovations more generally, such as the role of price and quality (Brécard et al., 2009; Brouhle and Khanna, 2012), and overcoming potentially negative user attitudes (Bhate and Lawler, 1997). ‘Customer relationships’ describe the relationship that a business has with customers, such as whether a CSA technological innovation provider will pursue a ‘one-off’ sale of a product, or whether further ‘after-sale’ services are offered. This building block also articulates the extent to which customers and end-users provide input to development or co-creation efforts with CSA technological innovation providers. How a firm links to end-users and creates a relationship with them is important for BMfSI (Boons and Lüdeke-Freund, 2013), as this is how user behaviour is influenced. For a BMfSI the customer relationship should encourage wider sustainability actions and could also act to minimise rebound effects. More widely, stakeholder engagement during the development process can increase the market acceptance of sustainable innovations, signalling that the customer interface should be broader than just the sales of the product (Byrne and Polonsky, 2001; Carrillo-Hermosilla et al., 2010). For BMfCSATI, the customer interface should encourage optimum usage of the innovation and seek to encourage wider action for CSA by end-users. ‘Channels’ link a business to its customers. Having a well-developed channel is a requirement for the successful diffusion of CSA technological innovations. This is highlighted as a critical aspect of BMfSI (Boons and Lüdeke-Freund, 2013), as the channel includes the marketing and awareness raising activities and strategies. Information provision is a key factor in the success of sustainable innovations (Lee et al., 2006), meaning the channel operation should ensure that CSA technological innovations have high awareness rates, and that CSA technological innovation providers ensure enough information is provided to the market. ‘Key resources’ are required to provide any product or service; this factor alone highlights this as a critical issue. However, the resources available to provide sustainable innovations will also effect price and quality (Brouhle and Khanna, 2012). ‘Key partners’ are needed as firms do not operate in isolation. Other actors often hold the resources required for delivery of the value proposition. The ‘delivery chain’, highlighted by Jabbour (2008) and Jabbour et al. (2013), confirms the importance of this block for sustainable innovation. Further, connections to wider networks could enhance the probability of a supportive regulatory environment (Ceschin, 2013). ‘Revenue streams’: BMfSI should seek to develop more innovative revenue models, including shifting towards pricing on ‘jobs done’ rather than per product (Boons and Lüdeke-Freund, 2013). This enhances the opportunity for de-materialising production and consumption. The ‘cost structure’ should be minimised, maximising the chance for profit. For BMfCSATIs the revenue model should at least allow the firm to be economically viable, ensuring that their products are priced competitively, as price is still a key consideration for end-users as with conventional innovations (Brécard et al., 2009; Brouhle and Khanna, 2012). A review of BMfSI literature and wider research on the key success factors for sustainable innovations indicates that the above noted business model elements are critical for the success of technological innovations, and in this case technological innovations for CSA. In Figure 1, these critical issues are mapped onto the BMC (Osterwalder and Pigneur, 2009).

International Food and Agribusiness Management Review

Long et al.

Volume 20, Issue 1, 2017

3. Methods Research approach/methods Data for this research was collected through the â&#x20AC;&#x2DC;CSA Boosterâ&#x20AC;&#x2122;, a European project funded by Climate-KIC. As part of the CSA Booster, data was collected that explored perceived barriers to the diffusion of CSA technological innovations as well as what would be required within the market for increased adoption and diffusion of CSA technological innovations. Data was collected from both CSA technological innovations providers as well as potential end-users. Data collection was undertaken with partners in the Netherlands, France, Italy and Switzerland. As CSA and related CSA technological innovations are a relatively new phenomenon, with little previous research conducted within developed world contexts such as Europe, the research took an exploratory approach. As such, the theoretical framework (Figure 1) utilised wider knowledge of a more generic and normative nature. This theoretical framework was then tested with empirical data collected within the context of CSA technological innovation diffusion to test the applicability of previous knowledge and to identify those business model elements most critical to the successful adoption and diffusion of technological innovations. Due to this exploratory approach, a qualitative stance was adopted. Semi-structured interviews with key informants were chosen as the data collection method, as this enabled rich and in-depth data to be gathered on factors acting as barriers to the diffusion of CSA technological innovations as well as what needed to improve to increase the diffusion of these innovations through the market. Both innovation providers as well as potential users of CSA technological innovation providers were included within the empirical sample. This allowed data to be collected from both those firms whose business models are the target of this research, as well as potential users within their target market. The potential users included actors such as farmersâ&#x20AC;&#x2122; associations, consumer goods companies and retailers, allowing the inclusion of perspectives from across agro-food chains. This adds value, as an awareness of how innovations will be used and integrated by users is noted as an important factor for innovation and the business models used to diffuse them (Stubbs and Cocklin, 2008). The user perspective also provided an external view on the current activities and operations of CSA technological innovation providers, enhancing the validity of the data and results. Financial actors were also included in the empirical sample due to their role in financing to both CSA technological innovation providers and users. Table 1 and 2 provide an overview of the data sources. Interviews used a semi-structured format. The questions were designed to identify key barriers faced, and what would be required within the market for an increase in the adoption and diffusion of CSA technological innovations. All project partners performed interviews, enhancing the geographical range and diversity of the data, using a standardised questionnaire protocol and recorded into a standardised template. Analysis of the data involved thematic coding, where typical and frequent answers were coded, as well as any data that was critical to the answering of the research question. Coding went through several iterations to enhance internal consistency within the themes developed. Once complete, a range of codes and categories were produced identifying key themes or critical issues. These were then applied to the BMfCSATI framework developed in Figure 1, which highlights which aspects of the business model are impacted. An overview of the conceptual and analytical process followed by the research is shown in Figure 2.

International Food and Agribusiness Management Review

Long et al.

Volume 20, Issue 1, 2017

Table 1. Climate-smart agriculture technological innovation provider data sources/participants.1 A1 A2 A3

A4 A5 A6 A7 A8 A9 A10 1

CSA technological innovation providers

Interview country

Provides software and new tools, managed through the internet, in order to optimise yields and quality of grapes in vineyards. The technology combines two competencies: remote sensing and agronomic models to develop precision farming and integration of satellite data in agronomic models. New technology is represented by the new certification service and related tool for communication. This approach can assist and improve products and processes sustainability. Biogas plants trading within Swiss emissions scheme and improve the quality of fertilizer. Feed additive that can reduce GHG emissions from ruminants. LED farming/urban farming. Use water treatment plant by-products to produce fertiliser and pig feed additives. Water salinity regulator. Remote sensing technologies for water. Precision irrigation systems that integrates meteorological data.

France France Italy

Switzerland Switzerland the Netherlands the Netherlands the Netherlands the Netherlands the Netherlands

CSA = climate-smart agriculture, GHG = greenhouse gas.

Table 2. Climate-smart agriculture technological innovation user data sources/participants. B1 B2 B3 B4 B5 B6 B7 B8 B9 B10 B11 B12 B13 B14 B15 B16

Users

Interview country

Financial services, with interests in agro-food chains. Dairy products company. European-level farmers’ association and trade body for agro-cooperatives. Farmers association. Sugar processor. Consumer goods provider. Agro-chain investor. Vendor for livestock feed products. Financial and industrial actor of French oleaginous and protein seeds industry. Potato farmers’ association. Farmers association. Support organisation for clean/pro-environmental farming techniques. National retailer. Global ‘snacks’ provider. Supermarket. Chocolate provider.

the Netherlands the Netherlands the Netherlands the Netherlands the Netherlands the Netherlands France France France Italy Italy Switzerland Switzerland Switzerland Switzerland Switzerland

4. Results Empirical data collected through the interviews will now be examined through the BMfCSATI critical issues framework, and compared to the critical issues that emerged during the literature review (Figure 1). Each aspect of the framework will be examined in turn and illustrated with examples from the data. During the data analysis it became clear that some key factors were relevant to more than one building block of the BMfCSATIs framework. As data was collected from both the supply and demand perspectives, the coding was carried out separately. This has the advantage of allowing the different perspectives to be compared and contrasted, before identifying a combined set of critical issues.

International Food and Agribusiness Management Review

Long et al.

Volume 20, Issue 1, 2017 1. Literature review

2. Theoretical framework

3. Empirical data Barriers and factors that could enhance diffusion from providers and users

Critical issues from previous research

4. Coding and analysis of key factors into BMCs Users Providers

5. Identification of critical issues for BMfCSATIs

Figure 2. Overview of conceptual and analytical process of the research. BMC = business model canvas, BMfCSATI = business models for climate-smart agriculture technological innovation. Climate-smart agriculture technological innovation provider perspective Value proposition – the value proposition was identified as a critical issue for BMfCSATIs, as it highlights the value of a technological innovation to the potential user. This has a large impact on marketing and sales, and in turn on adoption and diffusion. CSA technological innovation providers noted difficulties in terms of proving the value and demonstrating the impact of their CSATIs. As these related to the role of the value proposition, these factors were located in the value proposition of the framework. [The] main problem was convincing potential customers that it works, since it’s a new technology. (A7) Convincing customers is a hard one. We do have small-scale pilot and demonstration projects at farmers’ places but [it is] still hard to convince the actual customers of buying the product. The risk for the farmer is high if it doesn’t work. It depends on the local circumstances, but mostly the outroll phase is the complicated one, it goes slowly. (A6) An enhanced ability to conduct impact analysis and verification, as well as the creation of ‘CSA’ certification was noted as a change that could enhance adoption and diffusion, i.e. that these actions would boost the adoption and diffusion of CSA technological innovations. These factors increase value, by highlighting the potential impacts of the innovations, and so were located within the value proposition block. Scientific verification, which is cheap and a support service for research activities, would be very useful. (A5) Moreover, linking to potential customers ... could be a real asset. (A10) Channels – CSA technological innovation providers noted difficulties in accessing customers. This was relevant to the ‘channel’ building block, as this describes how firms connect with their customers and endusers; in the case of CSA technological innovation providers: Clearly reaching the customer is the biggest issue at the moment. (A10)

International Food and Agribusiness Management Review

Long et al.

Volume 20, Issue 1, 2017

Correspondingly, CSA technological innovation providers noted that improved access to customers would enhance their ability to diffuse their innovations. This clearly linked to the role of the channel block in a business model. We are looking for a European network to increase adoption of this technology in different regions in Europe. We would love to have better access to knowledge institutions, retailers and customers. (A6) Key resources – it was found that CSA technological innovation providers found it difficult to access capital and investment. This barrier was coded into the key resources block, as capital and finance can be conceptualised as a key resource required to provide the value proposition. For instance, without this, CSA technological innovation providers lack the resources for suitable marketing campaigns or investment into customer relations development. We don’t ask for investors. The growth has been funded by clients, but we faced barriers; we are an innovative firm, not mature, so we can’t give guarantees for investors. But at this step we want to control the development of our own technology. (A1) Also the investment and access to investors is something they could really use. (A9) Improved access to investment and finance was also articulated as something that would enhance the diffusion and adoption of CSA technological innovations. Key partners – regulatory and policy difficulties were noted as a barrier and were included within this building block, as they relate to poor access to wider networks, including those related to policy and lobbying. Without access to actors with influence or understanding of policy, CSA technological innovation providers noted that their technological innovation were at a disadvantage. On the policy side it is also very interesting. It is hard to get water boards and the ministry along … the policy environment is not made for these kinds of technologies. (A8) CSA technological innovation providers noted improved access to wider networks would enhance their ability to diffuse the technological innovations. We are looking for a European network to increase adoption of this technology in different regions in Europe. We would love to have better access to knowledge institutions, retailers and customers. (A6) Cost structure – the data indicated that CSA technological innovations providers felt that their CSA technological innovations were too expensive and had uncompetitive return of investment (ROI) periods. Whilst this would impact their value proposition, high costs can be traced most directly to cost structures, a specific block in the framework. Currently we are charging 20,000 euros for a device but we want to bring this down to 6,000 euros since it clearly was too expensive. (A9) Figure 3 presents an overview of the results for the CSA technological innovation providers, highlighting the specific barriers and changes that could boost adoption and diffusion. These factors are plotted into relevant blocks within the business model framework.

International Food and Agribusiness Management Review

Long et al.

Key partners

Volume 20, Issue 1, 2017

Key activities

Barriers: • Lack supporting partnerships. Market demands: • Access to regulatory networks.

Value proposition

Customer relationships Customer segments

Barriers: • Convincing customers of worth. • Providing impact. Key resources Barriers: • Lack of capital/ investment (i.e. for marketing, expansion etc.). • Lack of market intelligence.

Market demands: Channels • Assessment and verification of CSATI Barriers: impacts. • Reaching/accessing customers. Market demands: • Access to customers.

Market demands: • Access to capital and investment. • Intelligence on market. Cost structure

Revenue streams

Barriers: • Too expensive. • Overly long pay-back periods.

Figure 3. Critical issues for business models for climate-smart agriculture technological innovations provider perspective results summary. CSATI = climate-smart agriculture technological innovation. Climate-smart agriculture technological innovation user perspective Value proposition – the customers and user perspective indicated a lack of verified impact or proof that CSA technological innovations would deliver as advertised. Through the coding, this was felt to highlight a deficiency with regard to the value proposition, as end-users were unsure or unconvinced of the technological innovations actual value. Technologies should have a proven impact, so farmers are convinced to use it. (B2) Information on the economic impacts of the technology should be emphasized to increase the adoption and diffusion of CSA technological innovations. For instance, one potential user noted: Don’t just focus on climate impacts but actually economic impacts, cost benefit analysis. (B4) These factors together were felt to correspond well to the value proposition aspect of the business model framework, and in the empirical case, can be considered a critical issue for BMfCSATIs. Channels – CSA technological innovation users highlighted they had too little information of available CSA technological innovations, with no knowledge of, or access to, providers. Not enough knowledge about such technologies. (B14) Changes that would increase adoption and diffusion mirrored the barriers, and included a desire for clear and ‘user friendly’ information. It’s more a language and communication issue. Talk to farmers about how the growing season is this year, don’t talk about climate change. It’s all in the language. (B4) International Food and Agribusiness Management Review

Long et al.

Volume 20, Issue 1, 2017

Make it connect with the daily business on the farm. Start with climate adaptation; as soon as you start with climate mitigation it becomes a ‘far away story’. Adaptation is more realistic to farmers, more closely linked to their daily lives, and easier to explain. (B4) Customer relationships – data indicated that users thought it was difficult to transfer knowledge to farmers. Also the knowledge of how to use the technologies, if there are any, is difficult to transfer to the farmers. (B16) Greater user involvement in the research, development and design process, through closer links with CSA technological innovation providers was noted as likely to improve CSA technological innovations for users, enhancing adoption and diffusion. Ask the farmers what they need, then do the research. Demand driven research and development. (B4) Also do you research together with farmers; only then will you be able to have an impact. (B4) These factors were coded into the customer relation block as this articulates how a business engages with its customers (and where information exchange is likely to take place). This could be in terms of the amount of information, or at what stage customer engagement begins. Cost structure – users identified similar difficulties to the providers in that they saw CSA technological innovations as generally too expensive (compared to existing products), with overly long return on investment periods. As with the similar response by providers, this was coded into the cost structure block. Many technologies have long payoff times and do not fulfil internal payoff criteria. (B13) Often costs. For instance, with the drip [irrigation] system: if you have a lot of acres, you will need a lot of rubber. Cost can be an issue then. [It is] only beneficial if you have the right conditions. In the end it is a cost-benefit analysis. (B4) Figure 4 provides an overview of the barriers and factors that could enhance future adoption and diffusion noted by the respondents. These factors are positioned into the relevant areas of the business model framework.

5. Discussion This research provides two contributions. Firstly, by mapping normative and generic critical issues for BMfSI on to the BMC, we developed a theoretical framework for critical issues for BMfCSATIs. Second, with empirical data we have been able to explore this theoretical framework and highlight critical business model issues for the diffusion of CSA technological innovations. Whilst previous research has identified factors that hinder the adoption and diffusion of CSA technological innovations factors, the novelty of this research is the application of these factors to the business models of CSA technological innovation providers. In this way, we identify critical issues for technological innovation diffusion in specific areas of the business model. By applying the barriers and factors that could boost the adoption and diffusion of CSA technological innovations to a business model framework, we are able to identify critical business model issues and propose business model innovations that could boost adoption and diffusion. As data was collected from both the providers and potential users of CSA technological innovations, we are also able to compare supply and demand perspectives of the current BMfCSATIs, and highlight potential deficiencies of CSA technological innovation providers’ business models, from two perspectives. This is an interesting result and further contribution, as it shows that perceived deficiencies of the current business models of CSA technological innovation providers are not mutually recognised by the two groups.

International Food and Agribusiness Management Review

Long et al. Key partners

Volume 20, Issue 1, 2017 Key activities

Key resources

Value proposition

Customer relationships Customer segments

Barriers: • Lack of verified information on CSATI impacts.. • Poor articulation (language) of value of CSATIs..

Market demands: • Training and advice on the use of CSATIs (i.e. after sale services). • Input to R&D efforts of CSATI producers.

Market demands: • CSATIs with glear, verified impact.

Channels Barriers: • No knowledge of available CSATIs. • Obstructive terminology. Market demands: • Access to, and information on, CSATIs.

Revenue streams

Cost structure Barriers: • Price/investment costs too high.

Figure 4. Critical issues for business models for climate-smart agriculture technological innovations user perspective results summary. Both groups identified problems and potential changes concerning the value proposition, channels and cost structures. But CSA technological innovation providers identified critical issues for key partners and key resources, which were not identified by potential users. This may not be surprising, as these are internal aspects of a business model, and not easily recognisable to customers or users. However, this is not the case with the deficiencies noted by potential users with regards to customer relationships, which were not recognised by the innovation providers. Figures 2 and 3 highlight graphically the critical issues for BMfCSATIs for both the user and provider groups, whilst Figure 5 provides a synthesised BMC, providing an overview of critical issues for BMfCSATI diffusion. Assessment of critical issues for business models for climate-smart agriculture technological innovations and previous knowledge A crucial question is what changes or innovations to these business models would increase diffusion rates of CSA technological innovations? This question allows this research to connect to the wider literature on business models and innovation, which often seeks to understand how business models innovations can enhance the diffusion of innovations (Baden-Fuller and Morgan, 2010). Our finding is that the current business models of CSA technological innovation providers, within our sample at least, do not currently appear to be designed or operating in an optimal way to diffusion CSA technological innovations. This lends support to assertions within the literature that appropriate business models are required for the success of innovations (Bohnsack et al., 2014; Boons et al., 2013; Teece, 2010). The BMfSI literature highlights several critical issues, which are reflected well in the results. One critical issue includes the need for low initial margins, to allow competition with establish firms (Chesbrough, 2010). Current CSA technological innovations were identified as overpriced and as having overly long ROI periods by both the innovation providers and potential users; this indicates that either margins are too high, or that the wider cost model of innovation providers is not optimised. This impacts on the value proposition through the need to provide proof or verification, which is likely to be exacerbated by the perceived high

International Food and Agribusiness Management Review

Long et al. Key partners

Volume 20, Issue 1, 2017 Customer relationships Customer segments

Key activities

Value proposition

Key resources

• Compelling value • Provision of ‘after proposition, consistent sale’ services, such as with principles of training and advice. CSA. • Cooperation with user • Clearly articulated in research and economic value/ development. impact of CSATIs. • Verifiable impact. Channels

• Create links to wider networks relevant to CSA, including policymakers.

• Access to suitable levels of capital and investment. • Access to market intelligence on CSATI users.

• Good connections between CSATI users and providers. • Effective CSATI awareness raising, through accessible terminology.

Revenue streams

Cost structure Barriers: • Price/investment costs too high.

Only identified by producers

Identified by users and providers

Only identified by users

Figure 5. Aggregated critical issues for business models for climate-smart agriculture technological innovations. price. Such factors correspond to the importance of price and quality identified within literature as important for the success of sustainable innovations more widely (Brécard et al., 2009; Brouhle and Khanna, 2012). A second critical issue concerns how customers will use the innovation (Teece, 2010) and the importance of integrating stakeholder perspectives, including those of customers, in the research and development process (Byrne and Polonsky, 2001; Carrillo-Hermosilla et al., 2010). Previous research also emphasises the importance of ease of adoption (Kemp and Volpi, 2008). Neither of these factors was recognised by the innovation providers, but were highlighted as factors that could increase diffusion by the potential users. This was coded into the customer relationship block. This discrepancy is important, as the lack of recognition of this factor by the innovation providers means that this critical issue is unlikely to be acted upon and integrated into future iterations of their business model. A third critical issue was located within the channel block and relate well to the previously identified importance of links to customers (Boons and Lüdeke-Freund, 2013), and to the significance of information for the success of sustainable innovations (Lee et al., 2006). Access to customers, or awareness of CSA technological innovation, was recognised by both sets of respondents. However, the importance of accessible information provision to the market was only noted by the potential users, highlighting a further area where innovation providers are unaware of a deficiency and critical issue. The need for a supportive regulatory environment for sustainable innovations highlighted by Nair and Paulose (2014), was found to be a critical issue in this case. A perception existed that CSA technological innovations were disadvantaged by the current policy environment. This was coded into the key partners block, as it reflects a lack of access or influence in the wider networks that can influence the market. A move away from per product pricing to ‘jobs done’ pricing is advocated in the BMfSI literature (Boons and Lüdeke-Freund, 2013). This factor was not recognised within our results by either providers or users. International Food and Agribusiness Management Review

Long et al.

Volume 20, Issue 1, 2017

This could be due to the more normative nature of the issue within the literature, compared to the more practical issues faced within the empirical sample. Equally, some CSA technological innovation providers may be charging via per ‘jobs done’, but neither providers nor users see this as a barrier or a market demand, meaning it was not uncovered by our questioning. Implications and innovations for business models for climate-smart agricultural technological innovations Building on the normative recommendations of the literature, it is possible to take each building block of the BMfCSATIs and the associated critical issues, and identify innovations that could encourage greater adoption and diffusion of CSA technological innovations. The value proposition building block for BMfCSATIs involves several critical issues. These include problems demonstrating impact and proving value, and ensuring that the economic value of CSA technological innovations was clearly articulated. Linked to the value proposition is the cost structure, with both users and providers noting that CSA technological innovations were too expensive with overly long return on investment periods. In order for BMfCSATIs to be enhanced, the value proposition must be made more compelling. This will involve reducing the price of CSA technological innovations, which may mean alterations to the cost structure, hence also reducing their ROI period, as well as ensuring that impacts of CSATIs are assessed to give users confidence. Loock (2012) notes that the value proposition should contain either the lowest price, best service or best technology. The customer relationship building block also contained several critical issues, including a demand for more ‘after sale’ services, as well as interaction with users during the development phase of the innovations. Business model innovation here should focus on the development of complementary services, such as training or education programs. To satisfy demands for user involvement in design, consideration and integration of principles such as user-centred design and/or open innovation models may improve BMfCSATIs (Abras et al., 2004). The lack of access to finance and capital, within the key resources block, could be addressed through moderation of the ‘key actors’ area of the business model. A lack of access to capital or investors may reflect poor links to wider networks including policymakers or venture capitalist. These difficulties will require different innovations within business models, but are all critical issues. Access to capital or investors as well as links to policymakers indicate business models that are too closed; BMfCSATIs need to open up (Huizingh, 2011) and create stronger links with wider stakeholders. Assessment of the business model perspective The BMfCSATIs framework, based on a synthesis of BMfSI and BMC, performed well in highlighting critical issues for the adoption and diffusion of CSA technological innovations and how they interact with the business models of innovation providers. This approach followed the third stream of business model research as it investigated how innovations and their associated business models compete in the market, impacting innovation adoption and diffusions (Chesbrough, 2010; Wirtz, 2011). We have identified key organisational level business model factors, peering into the ‘black box’ of firms, noted as somewhat neglected in the literature to date (Boons and Lüdeke-Freund, 2013). The sample enabled us to include several interorganisational elements and how these relate to the business model. The framework we developed through a review of literature on business models, sustainable innovations and their interaction, successfully identified all but one of the critical issues highlighted in the data; this can be observed by comparing Figure 1 and 4. The critical issue which did not emerge from the coding, revenue streams concerning ‘jobs done’ pricing, may not have been a critical issue purely because it may operate already within this market, and not be perceived to be a barrier. International Food and Agribusiness Management Review

Long et al.

Volume 20, Issue 1, 2017

Although business model innovations offer avenues for improvement in the adoption and diffusion of CSA technological innovations (as outlined in the previous section), external assistance will be required where external influences are present, which are beyond the control of the innovator. These may include low consumer demand for CSA products as well as more general policy and regulatory issues that would be unlikely to be overcome via business model innovations alone. Indeed, it is highlighted by (Ceschin, 2013) that sustainable innovations may require financial and regulatory protection during their infancy, which is an issue that must be addressed by external stakeholders,

6. Conclusions This article has highlighted and explored critical issues for BMfCSATIs. By identifying both barriers and factors that could improve the adoption and diffusion of CSA technological innovations, and by locating where in the business models of innovation providers these barriers and factors operate, we have examined how business models impact the adoption and diffusion of technological innovations. Business model deficiencies were identified, as well as remedial business model innovations. As current BMfCSATIs are non-optimal, they can be seen to be one barrier to the uptake of CSA technological innovation in Europe. That said, CSA technological innovations represent a relatively new market area, meaning many innovation providers are small and young firms likely to be subjected to many of the barriers inherent within start-up firms. The central practical implications include the need to develop innovation provider business models through both internal business model innovation and external supporting actions, such as the provision of advice or access to finance. By conducting this research, we have contributed to a wider literature on business models and their relationship to innovations and sustainable innovations. There was broad consistency between the generic and normative factors identified in the construction of the theoretical framework, and the empirical findings. Whilst we feel that the framework performed satisfactorily, the data used to populate it represented a relatively small sample. This opens up opportunities in the future to supplement the studyâ&#x20AC;&#x2122;s exploratory approach. The BMfCSATIs would benefit from further development, possibly through the use of in-depth case studies or a broader sample in order to validate the results.

Acknowledgements The authors are grateful for the financial support provided by Climate-KIC that enabled the research upon which this paper is based.

References Abras, C., D. Maloney-Krichmar and J. Preece. 2004. User-centered design. In Berkshire Encyclopedia of Human-Computer Interaction. Edited by B.W. Sims. Berkshire Publishing Group, Great Barrington, MA, USA. Baden-Fuller, C. and M.S. Morgan. 2010. Business models as models. Long Range Planning 43: 156-171. Benijts, T. 2014. A business sustainability model for government corporations. A Belgian case study. Business Strategy and the Environment 23: 204-216. Bhate, S. and K. Lawler. 1997. Environmentally friendly products: factors that influence their adoption. Technovation 17: 457-465. Bogdanski, A. 2012. Integrated food-energy systems for climate-smart agriculture. Agriculture and Food Security 1: 9. Bohnsack, R., J. Pinkse and A. Kol. 2014. Business models for sustainable technologies: exploring business model evolution in the case of electric vehicles. Research Policy 43: 284-300. Boons, F. and F. LĂźdeke-Freund. 2013. Business models for sustainable innovation: state-of-the-art and steps towards a research agenda. Journal of Cleaner Production 45: 9-19. International Food and Agribusiness Management Review

Long et al.

Volume 20, Issue 1, 2017

Boons, F., C. Montalvo, J. Quist and M. Wagner. 2013. Sustainable innovation, business models and economic performance: an overview. Journal of Cleaner Production 45: 1-8. Brécard, D., B. Hlaimi, S. Lucas, Y. Perraudeau and F. Salladarré. 2009. Determinants of demand for green products: an application to eco-label demand for fish in Europe. Ecological Economics 69: 115-125. Brouhle, K. and M. Khanna. 2012. Determinants of participation versus consumption in the Nordic Swan eco-labeled market. Ecological Economics 73: 142-151. Byrne, M.R. and M.J. Polonsky. 2001. Impediments to consumer adoption of sustainable transportation. International Journal of Operations & Production Management 21: 1521-1538. Carrillo-Hermosilla, J., P. del Río and T. Könnölä. 2010. Diversity of eco-innovations: reflections from selected case studies. Journal of Cleaner Production 18: 1073-1083. Ceschin, F. 2013. Critical factors for implementing and diffusing sustainable product-service systems: insights from innovation studies and companies’ experiences. Journal of Cleaner Production 45: 74-88. Chesbrough, H. 2010. Business model innovation: opportunities and barriers. Long Range Planning 43: 354-363. Chesbrough, H., S. Ahern, M. Finn and S. Guerraz. 2006. Business models for technology in the developing world: the role of non-governmental organizations. California Management Review 48: 48-61. Chesbrough, H. and R.S. Rosenbloom. 2002. The role of the business model in capturing value from innovation: evidence from Xerox Corporation’s technology spin-off companies. Industrial and Corporate Change 11: 529-555. Ching, H.Y. and C. Fauvel. 2013. Criticisms, variations and experiences with the business model canvas. European Journal of Agriculture and Forestry Research 1: 26-37. Coumou, D., V. Petoukhov, S. Rahmstorf, S. Petri and H.J. Schellnhuber. 2014. Quasi-resonant circulation regimes and hemispheric synchronization of extreme weather in boreal summer. Proceedings of the National Academy of Sciences 11: 12331-12336. Doganova, L. and M. Eyquem-Renault. 2009. What do business models do?: Innovation devices in technology entrepreneurship. Research Policy 38: 1559-1570. European Institute of Innovation and Technology (EIT). 2014. Innovation Communities. European Institute of Innovation and Technology Report. Available at: http://eit.europa.eu/interact/contact-us. European Commission. 2014. Climate action: low carbon technologies. European Commision Report. Available at: http://tinyurl.com/3uowt2u. European Commission. 2014. Climate action: what is the EU doing? Climate action. Available at: http:// tinyurl.com/nkky3bu. Food and Agriculture Organization of the United Nations (FAO). 2010. ‘Climate-smart’ agriculture: policies, practices and financing for food security, adaptation and mitigation. Available at: http://tinyurl. com/65nfr7k. Food and Agriculture Organization of the United Nations (FAO). 2014. FAO success stories on climate-smart agriculture. Available at: http://tinyurl.com/n993xsz. Fettke, P. and P. Loos. 2007. Reference modeling for business systems analysis. Idea Group Publishing, Hershey, PA, USA. Geels, F.W. 2005. Processes and patterns in transitions and system innovations: refining the co-evolutionary multi-level perspective. Technological Forecasting and Social Change 72: 681-696. Hansen, E.G. F. Grosse-Dunker and R. Reichwald. 2009. Sustainability innovation cube – a framework to evaluate sustainability-oriented innovations. International Journal of Innovation Management 13: 683-713. Hanshaw, N. and A. Osterwalder. 2015. Why and how organizations around the world apply the business model canvas. Strategyzer, Zurich, Switzerland. Horbach, J., C. Rammer and K. Rennings. 2012. Determinants of eco-innovations by type of environmental impact – The role of regulatory push/pull, technology push and market pull. Ecological Economics 78: 112-122. Huizingh, E.K.R.E. 2011. Open innovation: state of the art and future perspectives. Technovation 31: 2-9. Iles, A. and A.N. Martin. 2013. Expanding bioplastics production: sustainable business innovation in the chemical industry. Journal of Cleaner Production 45: 38-49. International Food and Agribusiness Management Review

Long et al.

Volume 20, Issue 1, 2017

Jabbour, C.J.C. 2008. In the eye of the storm: exploring the introduction of environmental issues in the production function in Brazilian companies. International Journal of Production Research 48: 6315-6339. Jabbour, C.J.C., F.C.A. Santos, S.A. Fonseca and M.S. Nagano. 2013. Green teams: understanding their roles in the environmental management of companies located in Brazil. Journal of Cleaner Production 46: 58-66. Kemp, R. and M. Volpi. 2008. The diffusion of clean technologies: a review with suggestions for future diffusion analysis. Journal of Cleaner Production 16: S14-S21. Kurukulasuriya, P. and S. Rosenthal. 2003. Climate change and agriculture. Climate Change Series. World Bank Environment Department, Washington D.C., WA, USA. Available at: http://tinyurl.com/gwx639p. Larson, A. 2011. Sustainability, Innovation and entrepreneurship. Flat World Publishing, Washington D.C., WA, USA. Lee, J.J., K. Gemba and F. Kodama. 2006. Analyzing the innovation process for environmental performance improvement. Technological Forecasting and Social Change 73: 290-301. Lin, R.-J., K.-H. Tan and Y. Geng. 2013. Market demand, green product innovation, and firm performance: evidence from Vietnam motorcycle industry. Journal of Cleaner Production 40: 101-107. Loock, M. 2012. Going beyond best technology and lowest price: on renewable energy investors’ preference for service-driven business models. Energy Policy 40: 21-27. Montalvo, C. 2008. General wisdom concerning the factors affecting the adoption of cleaner technologies: a survey 1990-2007. Journal of Cleaner Production 16: S7-S13. Nair, S. and H. Paulose. 2014. Emergence of green business models: The case of algae biofuel for aviation. Energy Policy 65: 175-184. Nelson, G.C., M.W. Rosegrant, J. Koo, R. Robertson, T. Sulser, T. Zhu, C. Ringler, S. Msangi, A. Palazzo, M. Batka, M. Magalhaes, R. Valmonte-Santos, M. Ewing and D. Lee. 2009. Climate change: impacts on agriculture and costs of adaptation. International Food Policy Research Institute, Washington D.C., WA, USA. Available at: http://tinyurl.com/gw7jfdl. Osterwalder, A. and Y. Pigneur. 2009. Business Model Generation. Université de Lausanne, Lausanne, Switzerland. Osterwalder, A., Y. Pigneur and C.L. Tucci. 2005. Clarifying business models: origins, present, and the future of the concept. Communications of the Association for Information Systems 16: article 1. Pohle, G., P. Korsten and S. Ramamurthy. 2006. Component business models: making specialization real. IBM Business consulting services – strategy and change. IBM Institute for Business Value, New York, NY, USA. Available at: http://tinyurl.com/ywag7f. Seuring, S. and M. Müller. 2008. From a literature review to a conceptual framework for sustainable supply chain management. Journal of cleaner production 16: 1699-1710. Shafer, S.M., H.J. Smith and J.C. Linder. 2005. The power of business models. Business Horizons 48: 199-207. Sivertsson, O. and J. Tell. 2015. Barriers to business model innovation in Swedish agriculture. Sustainability 7: 1957-1969. Stubbs, W. and C. Cocklin. 2008. Conceptualizing a ‘sustainability business model’. Organization and Environment 21: 103-127. Teece, D.J. 2010. Business models, business strategy and innovation. Long Range Planning 43: 172-194. Trnka, M., R.P. Rotter, M. Ruiz-Ramos, K.C. Kersebaum, J.E. Olesen, Z. Zalud and M.A. Semenov. 2014. Adverse weather conditions for European wheat production will become more frequent with climate change. Nature Climate Change 4: 637-643. Wells, P. 2008. Alternative business models for a sustainable automotive industry. In Perspectives on Radical Changes to Sustainable Consumption and Production 1. System Innovation for Sustainability. Edited by A. Tukker, M. Charter, C. Vezzoli, E. Stø and M. M. Anderson. Greenleaf, Sheffield, UK, pp. 80-98. Wirtz, B.W. 2011. Business model management: design – instruments – success factors. Gabler Verlag, Wiesbaden, Germany.

International Food and Agribusiness Management Review

OPEN ACCESS International Food and Agribusiness Management Review Volume 20 Issue 1, 2017; DOI: 10.22434/IFAMR2015.0060 Received: 7 May 2015 / Accepted: 14 October 2016

Assessment of socio-economic configuration of value chains: a proposed analysis framework to facilitate integration of small rural producers with global agribusiness RESEARCH ARTICLE Miguel Arato a, Stijn Speelmanb, Joost Desseinb, and Guido van Huylenbroeckb aResearcher

and bProfessor at the Department of Agricultural Economics, Faculty of Bioscience Engineering, Ghent University, Coupure Links 653, 9000 Ghent, Belgium

Abstract Value chain analysis is an important tool to assess and enhance the performance of agribusiness. This paper analyzes the empirical application of a conceptual framework known as the Rural Web to evaluate the socioeconomic complexity of a specific agribusiness value chain. This can be used as a complementary approach to traditional value chain analysis. The proposed framework goes beyond linear descriptions of product flows and examines how supply chains are built, shaped and reproduced over time and space, while considering social, cultural, environmental and political aspects. The results demonstrate that the proposed framework is a suitable method for value chain analysis, principally for those whose early stages are based on small and medium-sized rural actors. The Rural Web analysis offers decision-makers a platform to identify key actors not traditionally considered in value chain analysis, as well as the social interrelationships that occur at different dimensions. It also enables the identification of corrective and preventive measures to enhance agribusiness value chains. Keywords: socio-economic analysis, agribusiness, value chain, Rural Web, sustainability, rural development, Mexico JEL code: Q01, Q13, R11, R23 Corresponding author: marato@global.t-bird.edu

ÂŠ 2016 Arato et al.

Arato et al.

Volume 20, Issue 1, 2017

1. Introduction Various authors have investigated the ways in which the relationship between small-scale producers in developing countries and agribusiness firms can act to enhance rural livelihoods (Blandon et al., 2009). At the same time, contemporary development policy prescriptions often place emphasis on the potential for closer integration of poor people or areas with global markets. In accordance with this perspective much of the literature has concentrated on exploring how firms and farms in developing countries can be integrated into global markets through value chains (Bolwig et al., 2010; Metzger et al., 2010; Trienekens, 2011). In addition, international experience has demonstrated that value chain analysis can be an important tool to enhance the performance of agricultural, food and fiber systems. It can help chain stakeholders and policymakers to identify corrective measures and to kick-start the development of areas and activities where the potential for growth has been identified (Da Silva and De Souza Filho, 2007). Many methods for value chain analysis have evolved in recent years, ranging from the more descriptive and qualitative to modelling and simulation studies (Fasse et al., 2009). However, working with small scale producers and principally with those from developing countries, generally with a series of socio-economic disadvantages, represents for the chain stakeholders additional complications and challenges that goes beyond the economic approach assessed by the traditional value chain analysis tools. Along with cost-reduction and improvement strategies, members of global value chains (sourcing channels, producers, distributors and final consumers) should also consider non-traditional aspects from the small-scale producers such as the social, cultural, environmental and political. As Marsden et al. (2010) noted, in order to fully understand the role and potential of chains, there is a need to move beyond descriptions of product flows, to examine how supply chains are built, shaped and reproduced over time and space. Authors like Tallontire et al. (2011) and Tallontire (2007) recognize the importance of analyzing the ‘horizontal’ along with the ‘vertical’ dimensions of governance in a value chain analysis. One potentially interesting approach is to examine regional food or natural resource chains from within, to focus upon the social relationships of trust and cooperation between the actors within the network with a view to identifying obstacles and opportunities (Jarosz, 2000; Peterson, 2013). Along the same line Block et al. (2008) propose a ‘value web’ approach that considers the different dimensions in which a value chain develops and explores the interactive and iterative relationships between the actors involved. This paper aims to go further in evaluating the socio-economic complexity of a value chain exploring the potential application of a methodology known as ‘Rural Web analysis’ (Van der Ploeg and Marsden, 2008) as a complementary approach to traditional-linear value chain analysis tools. We believe that by using the Rural Web the interconnections between six different dimensions (described below) of the value chain can be assessed, generating analytical insights that would remain hidden when applying a linear (producer to consumer) analysis of a value chain, as shown in Figure 1. Moreover, greater understanding can be achieved in relation to the strategic linkages – both between the actors involved in generating a product’s value and also with other actors who have an indirect effect on the agribusiness value chain. In order to demonstrate the appropriateness of the methodology for the above described purpose, we applied the Rural Web empirically to the case of Candelilla wax producers in a number of rural communities in the Chihuahuan desert in northern Mexico. The broad range of contextual and other complex conditions that are covered in the Rural Web make a case study the most suitable way to test the applicability of the Rural Web for value chain analysis (Yin, 2012). Furthermore, this rural web analysis was part of a broader evaluation of the potential of the Candelilla wax supply chain to foster rural development in the region (Arato, 2016). In the current article the potential of the tool is shown and several interesting outcomes for the Candelilla case are generated: it will increase common understanding of the rural reality (as perceived by the different actors involved); it can serve as a starting point to negotiate and resolve the identified differences in opinions, needs and expectations; it can identify possible pathways to foster efficient improvement strategies; as well as enabling the proper identification and management of possible risks to ensure the sustainability of the value chain.

International Food and Agribusiness Management Review

Arato et al.

Volume 20, Issue 1, 2017 Traditional value chain analysis

Economic (supplier-customer) linkages/product flow

Rural Web value chain analysis Multi-dimensional linkages D1 D6

Economic (supplier-customer) linkages/product flow

D1: Sustainability D3: Endogeneity D5: New institutional arrangements D2: Novelty production D4: Social capital D6: Market governance

Figure 1. Conceptualization of traditional and rural web value chain analysis. Adapted from Kaplinsky and Morris (2002) and Fasse et al. (2009). The representation of rural web value chain analysis is adapted from Van der Ploeg et al. (2008) and Marsden (2010).

2. Frameworks for value chain analysis Traditional analysis of value chains The field of value chain theory has received a lot of attention during the last decades (Bolwig, 2010; Fasse et al., 2009; Kaplinsky and Morris, 2002; Porter, 2011; Tallontire, 2007; Tallontire et al., 2011; Trienekens, 2011). According to Trienekens (2011) different disciplines have added to the development of value chain theory, including global value chain analysis; new institutional economics; supply chain management; and social network theory. Social network theory for example explains that value chains are shaped by concepts beyond economic considerations. Aspects like trust, reputation and power have a key impact on the structure and duration of value chains (Uzzi, 1997; Trienekens, 2011). In terms of methodologies to assess the evolution and performance of value chains, a common element to identify the different actors, activities, flows, inputs and outputs is the â&#x20AC;&#x2DC;mapping of value chainsâ&#x20AC;&#x2122;. The most common type of mapping is the vertical representation. This provides a linear description of product flows from input to output including the relationships among the involved actors, as well as the economic activities at each stage, as shown in Figure 1. Such type of mapping is often the basis for modeling, accounting and econometric exercises to evaluate the optimization and efficiency of value chains (Fasse et al., 2009; Kaplinsky and Morris, 2002). Other types of mapping, principally related to the social network theory, constitute a combination of vertical and horizontal mapping. During the last decades the vertical-horizontal mapping has gained popularity because of the incorporation of the social context. Such mixed mapping provides a visual representation of the connection with other value chains, as well as a description of the social relationships of trust and cooperation between groups and organizations with a view to identifying obstacles and opportunities (Borgatti and Li, 2009; Jarosz, 2000; Peterson, 2013; Tallontine et al., 2011; Trienekens, 2011). The current article relates to this literature and proposes the Rural Web analysis as a way to get a more holistic view on value chain performance.

International Food and Agribusiness Management Review

Arato et al.

Volume 20, Issue 1, 2017

The Rural Web The Rural Web, as described by Van der Ploeg et al. (2008), is an analytical tool that allows for a thorough exploration of the characteristics of specific localities, wider regional settings and development initiatives. It consolidates the large body of theoretical and empirical work on rural matters. In accordance with Horlings and Marsden (2012) a ‘Web’ could be defined as the relational system through which the human and the ecological components of a territory interact and intersect. Six dimensions, each of which highlight particular features of the web, can be distinguished (Marsden 2010; Van der Ploeg et al., 2008): ‘sustainability’, ‘novelty production’, ‘endogeneity’, ‘social capital’, ‘new institutional arrangements’ and the ‘governance of markets’. A series of potential applications of the Rural Web has already been identified (Messely et al., 2013): as a tool for comparative analysis of different development paths, within and between regions; as an approach to sustainable rural development; and as a diagnostic tool for exploring the potential limits of rural development patterns. This paper, however, is the first attempt to use it to evaluate a supply chain. On the one hand, this application is based on the recognized importance of the value chain as the locus of economic growth and development for small and medium sized rural producers and their communities (Arnold and Ruiz Pérez, 1996; Belcher et al., 2005; Belcher and Schreckenberg, 2007; Fisher and Dechaineaux, 1998; Marshall et al., 2006; Syampungani, 2009). On the other hand, it reflects the premise of Van der Ploeg et al. (2008) who stated that the performance of a regional economy, its comparative advantages, its competitiveness, innovativeness and sustainability could be explained by a functioning and comprehensive web. Contrary to the linear-traditional value chain analysis which generally include only those actors who participate actively in the product lifecycle (from seeding, collection and processing of raw materials to industrial processing, distribution and trading of final value added goods), the use of the Rural Web would allow to identify all the actors involved at a regional level of a value chain. The Rural Web analysis framework aims to provide insights from the actors in the product lifecycle, but also from those who are not actively involved in the process, such as universities, research institutes, and government representatives. These external actors do not provide a tangible value to the product; however they could have a significant impact on the value chain performance (Bitzer and Arts, 2013; Gregoratti, 2011; Tallontire et al., 2011). They could potentially limit its growth, and in some cases, hinder it completely by applying external impacts to the process.

3. Description of the study case The case study area is composed of rural communities from the Chihuahuan Desert, a region which extends over 450,000 km2 and includes part of the northern Mexican states of Chihuahua, Coahuila, Durango, Nuevo León, San Luis Potosí and Zacatecas (Schneider, 2009). The region is characterized by an arid and semi-arid ecosystem, with low levels of rain and extreme weather conditions which limit the potential for agricultural activity. As a consequence, the collection of Non Timber Forest Products is the main source of income for most families. One of these products is Candelilla. For about 3,000 families from this region the extraction and processing of this plant represents the main source of income. The commercialization of Candelilla is regulated by the Mexican secretary of environment and natural resources ‘SEMARNAT’ through its Mexican Official Norm: NOM-018-SEMARNAT-1999. The plant is collected from its wild environment and processed to extract a wax product known as ‘Cerote’, which is transformed into Candelilla wax through a refining process. The activity is mainly undertaken by men with sporadic support from the women. Most Candelilleros, as they are called, collect the plant from communitarian land properties. These are extensions of common land provided to a group of tenants, based on the Political Constitution of the Mexican United States (Mexican United States, 2015) and the internal regulations of the communitarian assembly. The tenants are entitled to undertake agricultural activities and to utilize the natural resources (including Candelilla and other wild species available). This is stated within the utilization permits issued by the Mexican authorities who regulate the rational use and preservation of the resources (CONAFOR, 2008). International Food and Agribusiness Management Review

Arato et al.

Volume 20, Issue 1, 2017

The production of Candelilla wax is a very old practice, and there are records of production volumes for industrial purposes back in the 1950s. For a number of reasons, the practice declined in popularity, and was even abandoned altogether in many places. For some years, production only took place in the Coahuila State and at much lower volumes. Recently, in the states of Nuevo Leon, Zacatecas, Durango, Chihuahua and San Luis Potosí the activity revived, mainly due to promotion by private firms and government institutions, which are encouraging rural development by developing natural resource based productive chains (Arato et al., 2014; CONAFOR, 2008). The value chain starts with the Candelilleros, who process the plant and trade it with local firms that then refine the product to produce the Candelilla wax. Due to its characteristics, Candelilla wax is a highly valued product for specialty applications in different industries, such as cosmetics, pharmaceuticals, the food industry, graphic arts and printing, among others. The main consumers of Candelilla wax are international markets in the USA, Europe and Asia, where 90% of the production volume is traded. The domestic market in Mexico accounts for less than 10% of the total volume produced. Mexican refinery companies are also participants in the Candelilla value chain, as they export the product to firms from the abovementioned markets (principally wholesale distributors and some large scale producers). Candelilla wax reaches final consumers in the form of a component for specialty consumption goods (Arato et al., 2014). The Candelilla wax value chain was the economic driver used in a rural development project encouraged by different actors (private firms, governmental representatives and civic organizations) to foster socio-economic development initiatives for Candelilleros (Arato et al., 2016). Due to the social, environmental and economic variables affecting the value chain as well as the variety of actors involved, the project leaders stressed their need for a methodology or tool that could provide a broader analysis of the social linkages and the relevant variables affecting the development of the value chain. This initiated the exploration of the application of the Rural Web as an analysis tool for value chains.

4. Methodology First the actors of the Candelilla wax supply chain were identified based on relevant literature (CONABIO, 2009; Schneider, 2009). They are presented in Table 1. Using this information, a plan for the interview process was established in order to obtain primary data about the perspectives of each type of actor. Primary data was obtained during a period of fieldwork in selected rural communities from three states in the Chihuahuan Desert (Nuevo León, Coahuila and Zacatecas), during July and August 2012 and during the same period in 2013. The communities were selected based on the importance of the production of Candelilla wax for their community and their experience of producing it. In order to select the rural producers, a general invitation was sent to all the Candelilleros from the selected communities. Those who accepted were interviewed. In total, interviews were conducted with 29 rural producers. In addition other types of actors were contacted and interviewed: 4 members from the private firm who initiated the rural development project, 2 members from a local university, 6 members from Forestry Governmental Agencies and Forestry Engineers and 1 member from a local governmental office. Overall 42 actors were interviewed. During the data collection the different aspects of the Rural Web were evaluated, using a semi-structured questionnaire. The interviews revolved around the six dimensions of the Rural Web. The respondents were presented with a number of statements related to each dimension, and they were asked to give their perspective on each of these statements using a 4-point scale reflecting their level of agreement with the statement. Each statement was developed based on the theoretical definition of each dimension (below described) and on previous knowledge from the region. The previous knowledge from the selected case study was obtained from local representatives and by working experience prior to the above mentioned fieldwork period. The local knowledge resulted relevant to formulate the statements according to the socio-economic conditions from the evaluated actors. It allowed to translate the theoretical definitions into applicable concepts. An overview of statements included in the questionnaire is given in the following paragraphs. The generally accepted definition of ‘Sustainability’ considers the existence of the social and ecological conditions necessary to support human life at a certain level of wellbeing for future generations (Earth International Food and Agribusiness Management Review

Arato et al.

Volume 20, Issue 1, 2017

Table 1. Actors interviewed during the Rural Web analysis. Actors

Characteristics

Candelilleros from Cuatrocienegas, Coahuila Candelilleros from Parras, Coahuila Candelilleros from Nuevo León Candelilleros from Zacatecas

This region represents the oldest producers of Candelilla wax. Contrary to the rest of the regions, these communities have shown continuous production over time. The Candelilleros from this region currently represent up to 40% of the total wax production in Mexico. This region is also a historic producer of Candelilla wax. These communities have shown continuous production over time. A newly re-activated producing region that, in the past, has shown discontinuous performance. However since the mid-2000s, it has shown an increasing production volume. This region, similar to Nuevo Leon, showed a negative and discontinuous performance during the 1980s and 1990s. However, since the mid-2000s the communities have been re-activating the production of Candelilla wax. CONAFOR is the government forestry agency responsible for promoting and encouraging preservation and development of sustainable commerce for natural resources. It also provides socio-economic organization programs to create productive chains through the strengthening of social organizations and institutional capabilities, as well as by providing training in the adequate use of forestry resources, with the purpose of generating employment and income. Forestry engineers are responsible for evaluating and analyzing the natural resources present in the rural communities, working with rural communities and CONAFOR to develop the exploitation permits in order to ensure the proper utilization of biodiversity (CONAFOR, 2012). Representatives from different local universities and research institutions were invited to participate in the evaluation process. The selection criteria were based on their experience in previous or current projects focused on the Candelilla wax or its production process. The representatives evaluated belong to The National Commission for the Knowledge and Use of Biodiversity ´CONABIO’; Universidad Autónoma de San Luis Potosí; Universidad Autónoma de Coahuila and; the research and development department from a local private firm (Multiceras) which is currently running different projects focused on the improvement of the production process as well as for different applications of the wax. The interviewed actors are involved in different projects focusing on research about the properties of the plant, diversification of applications and research into methods to make the production process more efficient. In this region there is significant integration between Candelilleros and municipal authorities. The representative of the municipal authority interviewed is a person with experience in the Candelilla collection process and he represents the interests of the rural collectors at local government level. This group includes collectors who are individuals located within the rural communities and generally work on a commission basis purchasing Cerote for different private firms. This group also includes members from different departments (Sales, Agribusiness and Social Responsibility) from a private firm that is recognized as the largest trader of Candelilla wax.

Forestry government agencies and forestry engineers

Universities and research institutions

Municipal authorities

Private firm

Council, 1994; Van der Ploeg et al., 2008). In accordance to this we developed a series of statements that would provide us with the insights and perspectives from the different actors interviewed, in relation to the following: (1) the continuity of the activity; (2) the extent to which the income generated by the activity is sufficient; (3) their opinion about the current economic revenue from the activity; (4) their opinion about the involvement of future generations in the activity; (5) their perception about future improvement in the activity; (6) whether they believe there is sufficient stock of the plant to continue processing; (7) their awareness and knowledge about recommendations for plant preservation; (8) their opinion about how involved the rural producers are in resource preservation measures; (9) their perception about the reforestation campaigns and their effectiveness. The respondents provided their perceptions on all of these aspects using a four point scale International Food and Agribusiness Management Review

Arato et al.

Volume 20, Issue 1, 2017

(for example for question 8, where the respondents were asked to indicate the degree of involvement of rural producers in resource preservation measures the scale ranged from ‘little involved’ to ‘highly involved’). The second dimension ‘Novelty Production’ refers to new insights, practices and artifacts developed to improve the process or the product. At the same time, it refers to deviation from the rule, distinguishing its results from the accumulated or expected knowledge (Marsden, 2010; Van der Ploeg et al., 2006). Therefore, under this dimension we evaluated to what extent changes or improvements are made by the Candelilleros: (1) in the plant collection process; (2) in terms of simplifying the wax production process; (3) to increase the production volume; and (4) for the preservation and reforestation of the plant. Other aspects included in the evaluation of this dimension were the level of diversification and multi-functionality currently taking place (Durand and Van Huylenbroeck, 2003; Ellis, 2000; Ellis and Biggs, 2001; Nemes, 2005). This involved asking about (5) the development of other initiatives for complementary activities (tourism, cultural events, traditions, other activities). For all these aspects, the respondents were asked to rate the extent to which these actions took place. In line with the premise stated by Van der Ploeg et al. (2008) about endogenous development which occurs when there is sufficient consensus about the goals of development and, consequently, about what can be considered as local resources and the value of local entities as resources, we included the following aspects in the evaluation of ‘Endogeneity’: (1) to what extent the Candelilleros depend on other entities (private firms, buyers, government) to collect and process Candelilla; (2) the level of dependency on others to obtain the processing equipment and supplies; (3) the dependency on others to learn the process and obtain training; (4) how the Candelilleros organized the development of community improvements (building of schools, roads, hospitals, etc.); and (5) how the Candelilleros attempt to increase the added value of the wax. The fourth dimension ‘Social Capital’, as defined by Fukuyama (1999) is an instantiated set of informal values or norms shared by the members of a group. It generally refers to trust and willingness to live by the norms of one’s own community and to punish those who do not (Bowles and Gintis, 2002; Durlauf, 2002). Therefore, within this dimension, we measured the ‘collective efficacy’ (Sampson et al., 1999) of the Candelilleros by analyzing their ability to engage in networks, cooperate and make use of social relationships for common benefit. We did this by formulating statements about: (1) how much cooperation exists between the Candelilleros from the same community (sharing knowledge, resources, support); (2) how organized the Candelilleros are to achieve common benefits; and (3) how good the relationship between Candelilleros from different communities is. The ability to achieve synergistic ‘win-win’ outcomes, as addressed by Barret et al. (2005), depends largely on the ‘New institutional arrangements’ that shape the incentives and constraints faced by human agents. In order to analyze this fifth dimension, we asked respondents to evaluate the institutional constellations that solve coordination problems and support cooperation among Candelilleros and different actors, such as: (1) local government institutions, (2) universities and research institutes; as well as (3) private firms. The final dimension considered is ‘Market Governance’. The capacity to control and strengthen markets, as well as to construct new ones, is related to the way in which a certain supply chain is organized, the distribution of the value created, and how the potential benefits of collective action are delivered. Candelilleros act as self-employed producers, working within their own territory and having the opportunity to decide whether to produce Candelilla wax or undertake any other complementary activity. To analyze ‘Market Governance’, we differentiated between two different market governance capacities: (A) the capacity of Candelilleros to control the market (sale price and offer), by evaluating: (1) their influence on the selling price of Candelilla wax; (2) the feasibility for Candelilleros to produce an additional quantity of Candelilla wax per month (10 kg, 30 kg, 60 kg extra). (B) The capacity of the other actors along the value chain to control the market (demand) by questioning: (3) the influence of private firms on the purchase price of Candelilla wax; (4) the influence of final users on the purchase price of Candelilla wax; and to analyze the demand volume of Candelilla wax

International Food and Agribusiness Management Review

Arato et al.

Volume 20, Issue 1, 2017

in the market, by (5) the feasibility for a private firm to purchase an extra quantity of 10, 30 and 60 kg of Candelilla wax per month, (considering that the average monthly production is 120 kg of wax per person). To analyze the results generated by the data collection process described above, we applied a mixed qualitativequantitative method. The ratings (from 1 to 4) obtained for each aspect were processed (calculating average scores per dimension) in order to map the results. This was undertaken to provide a picture of the general perception regarding each dimension, which was translated in the Rural Web chart (Figure 2), as explained in the results section. This analysis was complemented and interpreted using the additional information obtained from the interviews and the informal conversations.

5. Results Sustainability Sustainability was considered to be a critical dimension by all the actors interviewed. As described above the dimension considered, for example, the continuity of the activity and the preservation of the resource over time. The group of Candelilleros from the region of Cuatrocienegas and Nuevo Leรณn were more concerned about this dimension compared to the Candelilleros from the other regions. Other groups concerned with performance in terms of this dimension were the representatives of the national and local forestry authorities. Their concern related to the knowledge and application of the officially recommended preservation measures and about the effectiveness of the reforestation campaigns. Nevertheless, respondents were generally optimistic about the survival of the practice over the next five to ten years, and had positive expectations for improvements to the activity and its value chain in the future, as well as a positive perception regarding the adequacy of bio-stocks of the plant in the region. For most actors interviewed, the main concern identified related to continuity for the current collectors over a period longer than 15 years. This is logical considering that the average age of the Candelilleros ranges between 40 and 45 years. Moreover, another factor affecting continuity is the low number of young Candelilleros available to maintain the practice in the future.

Sustainability 4

Governance of markets

3 Endogeneity 2 1

Institutional arrangements

Novelty production

Social capital

Total average

Figure 2. Rural Web results. This figure is for illustrative purposes only. It was included to provide a visual reference about the total average of rates per dimension, as reported by the interviewed actors. The figure aims to facilitate the analysis of the dimensions and its relevance in accordance to the evaluated concepts. It does not intend to provide a quantitative analysis of the results.

International Food and Agribusiness Management Review

Arato et al.

Volume 20, Issue 1, 2017

Novelty production It was observed that novelties are mainly expected to take place in relation to the collection and production process. The types of novelty observed in the process vary according to the level of expertise and technological skills available. The concept of ‘Peasant Innovativeness’ (Oostindie and Broeckhuizen, 2008; Ventura and Milone, 2005) is reflected in this case in the small improvements developed by Candelilleros principally in relation to collection techniques and the production process. For example, using trucks or customized tools to collect and process the Candelilla plant can be considered as small improvements that save both time and effort. The actors interviewed from universities, research institutes and private firms are less optimistic concerning the level of improvements, basically because their expectations in terms of product and production efficiency relies on the application of advanced technology. Nevertheless, in this case, some technological improvements, such as new designs for processing equipment and furnaces, have been developed in recent years, principally focusing on improvements in production efficiency and working conditions for the Candelilleros. These changes are currently in place and were developed based on local research supported principally by the Mexican government (CONAFOR, 2008). Other research efforts were undertaken by the private firm Multiceras, which is currently working on three main objectives: to improve production efficiency and the production process and to search for new and improved product characteristics and applications (Multiceras, 2013). In order to increase the income of rural Candelilla collectors, two projects were developed by the Candelilleros from Cuatrocienegas, with support from the municipal authorities in cooperation with other national organizations. The projects involved the organization of workshops (in two different locations) to produce value added Candelilla wax. However, when the representatives from the private firms and research institutes were asked if they were familiar with these projects, they stated that there was limited information about the projects mentioned and, moreover, they lacked basic information on matters such as production capacity, the capabilities of the production equipment or even the specifications of their final product. Nevertheless, the representatives showed an interest in this project, because it would allow the Candelilleros to advance one further step in the value chain. Other improvement projects were encouraged by different institutions such as Universidad de Coahuila, which along with the governmental agricultural agency ‘SEMARNAT’ and ‘CONAFOR’ developed a project to increase the production efficiency and to improve the safety in working areas. Finally, in relation to multifunctionality, we could observe different activities being undertaken alongside agricultural-related activities, such as tourism and other cultural activities, principally in communities from Coahuila and Nuevo León1. Endogeneity The shared concerns with regard to this dimension relate to the general perception that Candelilleros are heavily dependent on external actors to obtain the technology and specialized skills to increase the added value of their product. Nevertheless, most of the interviewees recognize the capacity of Candelilleros to transfer their knowledge and production skills from generation to generation, as well as their internal organizational skills to generate shared benefits. This is certainly the case in the communities from Coahuila. This is a location where stronger collaboration exists between the Candelilleros and where the activity is more mature and extensive. Compared to the rest of the regions, Cuatrocienegas showed a higher level of organization between the community members to generate their own benefits and to provide added value to the product. At the time of this research,

1 The

local governments from the regions of Cuatrocienegas, and Parras in Coahuila and García and Mina from Nuevo León encourage different historical and eco-tourism attractions (http://tinyurl.com/zdg95xa; http://tinyurl.com/o87cequ; http://tinyurl.com/hwwpbqo).

International Food and Agribusiness Management Review

Arato et al.

Volume 20, Issue 1, 2017

there were some ongoing rural development projects through productive chains, encouraged by municipal and national support agencies (CONAFOR, 2012). Another peculiarity that distinguishes Cuatrocienegas is the larger presence of Candelilla wax buying private firms. This situation provides Candelilleros with greater negotiation power, which gives them more influence as a recognized group of producers. As already indicated, the local municipal authorities from Cuatrocienegas are closely involved in the interests and wellbeing of the Candelilleros as a group. Notwithstanding their organizational strength, the group of Candelilleros from Cuatrocienegas agreed that they are heavily dependent on third parties or government officials to provide them with production equipment and tools. When they were asked about the possible reasons for that dependency, they argued that for them it is more beneficial to keep receiving the tools from the buyers and save costs, as well as to avoid possible conflicts of interest by using common equipment and utilities. In general, of all the groups interviewed, the research institutes and the private firms were least optimistic about this dimension. Their main concerns related to the high level of dependency on external actors to obtain production equipment, the motivation to undertake the activity and the limited value added. Social capital All types of actors interviewed agreed on the high level of cooperation and support that exists between members of the same community (as shown in Figure 2). Rural communities are organized through community assemblies known as ‘Comisariado’, which are formed by the members of the community. The common activities developed in the community are first discussed between the members of the assembly and approved based on general acceptance. However, although there is a lot of cooperation between Candelilleros from the same community, limited cooperation was reported between members from different communities. The ‘Comisariado’ is the local institution responsible for promoting the common benefit of all the members of a given rural community, including those who are not Candelilleros but who still have a voice and a vote concerning the activities developed within the premises and the communities. We could observe situations where differences arise between members of the community, principally in terms of land use and the balance between other economic activities and the interests of those members not producing Candelilla. In these cases, the Comisariado would take the role of mediator between the different interests, establishing measures to negotiate the best possible outcomes for all the actors. On the other hand, some rural collectors mentioned cases where youngsters, who had left the community and moved to larger cities to work in factories or the construction industry, had returned after a couple of months and re-engaged in Candelilla collection. They argued that this occurs because in the city there are additional expenses such as rent and transport, and the cost of food tends to be much higher compared to the prices at the communitarian stores. These youngsters did not take such costs into consideration when making their decision to relocate. New institutional arrangements Detected as a main improvement opportunity, this dimension, as shown in Figure 2, was perceived most critical by all the respondents. The general concern is based on the discontinuity and the limited relationship between the Candelilleros and the local governments, universities and research institutes. The strongest institutional relationships were observed between the Candelilleros and the private firms, principally because of their working relationship which represents, for the Candelilleros, their main source of income and supply of production equipment. However, with the exception of the above mentioned case of Multiceras, in most cases the relationship with private firms is limited to the selling-purchasing process. As observed during the International Food and Agribusiness Management Review

Arato et al.

Volume 20, Issue 1, 2017

fieldwork period, Cuatrocienegas has a higher number of producers, compared to the rest of the regions, and attracts support from government and research institutes, both at local and national level, for improvement projects focused on infrastructure and social organization. This situation is not reflected in the rest of the Candelilla communities due to the lower number of producers. Market governance All the interviewees agreed that there is scope for improvement in terms of this dimension, principally for the market governance capacity of rural actors which is mainly determined by the influence of the members at later stages of the value chain. Regarding the two different market governance capacities we observed the following. Concerning the influence of the Candelilleros on the selling price of wax, which is the Candelilla wax in its simplest form, the rural producers from all the areas have a more positive perception, compared to the opinion of the rest of the actors interviewed. The general perception of the Candelilleros is that they can somehow influence the selling price by trading their product with the different private companies that purchase the Candelilla wax. However, the actors interviewed from research institutes, universities, government representatives and private firms believe that Candelilleros have limited influence on the selling price and state that the price is relates to more external factors such as global demand and variability in the selling price of competing and substitute natural wax products. In terms of the supply of Candelilla wax, we evaluated the feasibility for a Candelillero to produce an additional quantity of Candelilla wax on top of what he is currently producing. In the short term, and considering the current working conditions, available tools and technology, an additional 10 kg of Candelilla wax per month was perceived as feasible. However, perceptions were less optimistic concerning the feasibility of producing larger additional quantities (e.g. 30 kg extra). Producing an extra 60 kg was regarded as impracticable, under the current conditions; and respondents suggested that some improvements in the process should be made to achieve this. With respect to the demand, the average perception was that it was highly feasible for private firms to purchase an extra 10 kg or 30 kg. However, in terms of a scenario involving an increase in supply by 60 kg in the short term, the results varied considerably between the different actors interviewed. While the general perception among the Candelilleros was that the market can afford the purchase of this additional quantity, the rest of the actors – principally the private firms – regarded this scenario as less feasible, arguing that an extra 60 kg would represent a 50% rise in the total volume of Candelilla wax available in the global market, which would create the problem of overstocking. According to the comments from firm representatives, an increment in the available volume of wax should be accompanied by a marketing strategy in order to place the additional volume within new markets or increase the number of product applications, in order to create a balance between supply and demand.

6. Discussion The Rural Web Analysis framework As shown in the previous sections, the Rural Web analysis framework allowed us to gain insight into the social complexity and interaction between different members of a value chain. The framework enabled us to understand different perspectives on the potential of Candelilla production. The results obtained showed territory-specific peculiarities of the region and the economic activities that are based on the utilization of the local natural resource, as addressed by Van der Ploeg and Marsden (2008) and Marsden (2010). Identification of all the actors involved at a regional level is important for the complete assessment of a value chain. Linear-traditional value chain analysis usually considers only those actors who participate actively in the product lifecycle (from seeding, collection and processing of raw materials to industrial processing, distribution and trading of final value added goods). All actors considered in linear analysis interact in the process and provide a specific value to the product. On the other hand, the Rural Web analysis framework provides insights from the aforementioned ‘value adding’ actors, but also from those who are not actively International Food and Agribusiness Management Review

Arato et al.

Volume 20, Issue 1, 2017

involved in the process, such as universities, research institutes, and government representatives. These external actors do not provide a tangible value to the product; however they could have a significant impact on the value chain performance (Bitzer et al., 2013; Gregoratti, 2011; Tallontire et al., 2011). They could potentially limit its growth, and in some cases, hinder it completely by applying external impacts to the process As shown in the results section, the concerns from the different groups in terms of the dimensions evaluated were quite different: while dimensions such as Social capital and Novelty production obtained more optimistic perceptions, dimensions such as New institutional arrangements, Endogeneity and Governance of markets received less optimistic answers. However, the purpose of using the Rural Web is not to take actions to improve these perceptions, because positive changes in one dimension might have a negative effect on other dimensions. On the contrary, it is more about understanding the peculiarities of the situation and acting according to its characteristics. It is important to bear in mind that this is a snapshot of the perceptions obtained from current actors. Situations are dynamic and could be modified based on the common dialog and understanding of the actors involved. The analysis framework presented could facilitate the generation of a diagnostic relating to the potential risks and improvement opportunities that must be addressed from the actors’ perspective in order to develop a successful dialogue and the understanding to generate efficient development strategies. Findings from the case study Improvements in the socio-economic wellbeing of disadvantaged areas can best be brought about by recognizing and encouraging the collective resources of a territory itself (Ray, 2000). Therefore in this section, we analyze the main differences found in the perceptions of the actors interviewed. We link this with existing literature in order to provide specific recommendations and promote a common understanding of each one’s point of view. Such common understanding would form a baseline to identify risks and maximize strengths for each dimension in order to address further development strategies. In terms of Novelty production, given that there is a marked difference between the perception of Candelilleros compared to the rest of the actors interviewed, a common dialog must be encouraged between all parties in order to understand that novelties are largely a deviation from the rule and generally do not correspond with the knowledge accumulated to date (Van der Ploeg et al., 2006: 200). Such a dialog would allow them to identify the ‘Contextual knowledge’ (Oostindie and Broekhuizen, 2008) generated by the accumulation of technological capabilities and skills from each region. As observed, most improvements reported by the Candelilleros consist of small changes to production and collection techniques, generally on a territorial basis, which result in a steady but ongoing increase in benefits, while for the rest of the actors (principally private firms, research institutes and universities) the expected outcomes are linked more to improvements in efficiency and capacity building in relation to the production process. In this case, identifying the differences in concepts among all parties would enable the dissemination of knowledge throughout different territories (Oostindie and Broekhuizen, 2008). Another observation was that standardization and dissemination of new production and preservation techniques encouraged by private firms, governments and universities faced limitations due to the highly localized novelty production that exists within the rural communities. Shared understanding and communication would allow all parties to potentiate the possible outcomes from novelty production such as: improving resources, fine tuning, boundary shifts, and re-patterning resource use (Ventura and Millone, 2005). When applying development initiatives, it is important to consider the traditional territory-specific incentives that people deploy to regulate themselves, such as, for example: solidarity, reciprocity, reputation, personal pride, respect, retribution and vengeance (Bowles and Gintis, 2002; Gray et al., 2014). In the case of Social Capital and Endogeneity, the stakeholders involved should encourage the development of a ‘collective efficacy’ (Anderson and Jack, 2002; Bellandi, 2001; Sampson et al., 1999) in order to foster cooperation between communities within the same territory in a hands-on approach. International Food and Agribusiness Management Review

Arato et al.

Volume 20, Issue 1, 2017

As described above, the limited communication and interaction between communities complicates the implementation of territorial and regional development projects. In general, Candelilleros perceive themselves as an organized community that follows a territorial-individualistic working philosophy. However, from the different regions, it is only in Cuatrocienegas that sufficient organizational skills have been developed to create an impact group sufficient to generate common benefits on a regional basis. For the rest, the working relationships have been on an individualistic basis, considering ‘Community-specific’ needs. As identified by co-actors, this individualistic culture generates a lack of efficiency in the distribution of resources (such as equipment, tools, technical skills, etc.). In this case, cooperation between the members from different communities should be encouraged through working relationships based on ‘Trust’ (Bitzer et al., 2013; Bowles and Gintis, 2002; Durlaunf, 2002). Trust could act like a lubricant that, in this case, would enable a more efficient distribution of knowledge and resources throughout different territories (Fukuyama, 1999). In terms of Sustainability, the main concern from all those interviewed, is the low number of young Candelilleros who would undertake the activity in the future. For this reason, current improvement projects include the integration of youngsters in the production process. Awareness about the official recommendations for preservation is high, since most rural actors argued that they have received talks and training about it. However, when put into practice, in most cases the recommendations appeared inefficient, in view of the conditions where the activity is undertaken. Therefore, rural actors have come up with their own preservation techniques focusing on land distribution and collection patterns, alternating collection in order to let the plant re-grow. These collection schemes are respected by all rural members of the community and are regulated by the Comisariado for each rural community. In order to integrate the rest of the actors into the preservation measurements proposed by the Candelilleros, it is necessary to understand the constraints they experience in their daily activities, and from this, construct mechanisms to generate effective preservation of the natural resource. As suggested by Boettke et al. (2008), in order for formal institutions, to ‘stick’ with the regular working process, it must be mapped onto the informal rules. With respect to New Institutional Arrangements, a stronger interaction between all actors in the value chain should be encouraged, in order to maximize rural development opportunities. Candelilleros from most regions agreed about the limited relationship that exists with research institutes and Municipal authorities. Actors should migrate from the traditional style of support into a more active role (Shucksmith, 2010). The first step in establishing effective institutional arrangements is to establish, monitor and enforce rules. Since every region currently works on a territorial-individualistic basis, it is understood that some differences could be encountered between regions (Barret et al., 2001). Finally, in terms of Market Governance, non-rural actors agreed that Candelilleros have a certain influence on the selling price. However, this influence is limited within certain price boundaries, because the purchase price relates more to external factors such as global demand and variability in the selling price of competing and substitute wax products, such as Carnauba wax. In order for the Candelilleros to advance in the value chain, it is necessary to foster the conditions needed to meet the requirements for technical skills and equipment; and to improve working conditions. This could be accomplished by integrating all the members of the value chain within multi-institutional networks to encourage the active participation of producers, consumers, local institutions, NGOs and related organizations (Block et al., 2008; Marsden and Renting, 2003; Shucksmith, 2010; Tallontire, 2011; Ventura et al., 2008). Unfolding the Rural Web of the Candelilla wax value chain As described by Messely et al. (2013), regionalized rural development is grounded in, and driven by, complex sets of internal and external interactions, which shape the relative attractiveness and competitiveness of rural spaces economically, socially, culturally and environmentally. In this section, we reviewed the interaction flow of the actors analyzed within the dimensions described in order to ‘unfold’ (Marsden, 2010; Van der Ploeg and Marsden, 2008) the Rural Web of the Candelilla wax value chain and thereby to understand the development trajectory of the case under analysis, as shown in Figure 2. International Food and Agribusiness Management Review

Arato et al.

Volume 20, Issue 1, 2017

Based on the findings from the analyzed case study, the Rural Web flow begins with an interaction cell comprising the dimensions of Social capital, New institutional arrangements and Endogeneity. This first cell represents the main interaction block, where Social capital, as the key initiator, plays an important role in the development process because, as explained above, it represents the strongest dimension of the web due to the closed interaction and relationship between the rural producers from the same community (bonding social capital relations). Although the relationship with other communities (bridging capital relations) was rather weak, the relationship among the community members was strong enough to facilitate negotiations with the rest of the project members (linking social capital relations). In the analyzed case, the initial approach and negotiations were made between members from each community and the private firm. In line with the observed culture and traditions from the rural producers, the negotiations and agreements were performed according to the local interests. According to the comments from the interviewed, generally the agreements on price, volume, and other criteria from a given community might be different to those from a neighboring community, even when dealing with the same company and selling the same product. In this same cell Endogeneity acts as a lubricant for Social capital, principally because of the consensus that exists among the actors interviewed about the value of local resources and traditional know-how concerning the production of natural goods (i.e. Candelilla wax), as well as the observed capabilities to bring about common benefits and endogenous development (i.e. the community of Cuatrocienegas which demonstrated a higher level of organization as well as development initiatives). Influenced by the first interaction cell, Novelty production would be determined according to the relationship from the previous dimensions, which are defined by the members of the value chain and shaped according to their interests. As shown in this case, novelty varied according to the available resources and the objectives pursued by each group. Acting as a key lubricant and directly linked to Novelty production, is Market governance. This dimension serves as a base for the value chain, providing resources and income opportunities to the interacting actors. It must be said that, contrary to Social capital and Endogeneity which are somehow a constant element of this web, Market governance varies according to the economic activity encouraged. It also influences the New institutional arrangements, which will be configured according to the economic activity encouraged and the rest of the dimensions from the first interaction cell (i.e. the type of organizations interacting within the value chain varies according to the available resources, traditions, regulations, economic activity, and others). The relationship between the previously mentioned dimensions comprised a second interaction cell, represented in Figure 3 with a dotted line. The interactions between the dimensions analyzed, along with the trade and environmental regulations that frame the entire Rural Web, have a direct effect on Sustainability which, in turn, is determined according to the actors involved, the natural goods selected for utilization and their preservation, as well as the sector of the population available to process them. This interaction flow as a whole leads to rural development as an outcome. Challenges and limitations of the Rural Web as value chain analysis framework Based on our experienced using the Rural Web for the analysis of the Candelilla value chain, we agree with the observations from Messely et al. (2013) with regard to the need of a regional learning process as a prerequisite for its application in order to get familiar with its dimensions and functions. To minimize the limitations, and maximize the outcome from a Rural Web analysis, it is recommended that private firms receive support from social scientists, NGOs or local representatives with sufficient regional knowledge. Their support would facilitate a proper understanding of the communities, open dialog channels and would provide sufficient input to the stakeholders to guarantee the successful use of the Rural Web. During the case study, additional to the previously mentioned support, a valuable contribution from regional actors was received during the development of the questionnaire: providing important recommendations about local language, expressions, wording, and current paradigms present within the evaluated communities. International Food and Agribusiness Management Review

Arato et al.

Volume 20, Issue 1, 2017

Social capital (key initiator)

New institutional arrangements

Endogeneity (lubricant)

(Key lubricant)

Novelty production

Market governance

First interaction cell Second interaction cell Trade and environmental regulations

Sustainability

Rural development

Rural Web flow

Connection between dimensions Rural Web flow

Figure 3. Unfolding the Rural Web. In our experience, additional to the pre-requisite of regional knowledge, it is relevant for those interested to apply the Rural Web, to generate a preliminary evaluation about key concepts such as: (1) authority and hierarchy levels within the communities; (2) identification of local formal and informal leaders; (3) trust and perception from rural producers concerning the involved stakeholders and; (4) interest of rural producers to participate in the value chain, which for some might represent a change from another economic activity and learn new techniques and processes. The previous concepts were found relevant to secure the successful application of the Rural Web. However there might be more concepts that shall be considered, which opens an opportunity for further research work and analysis.

7. Conclusions The information generated by the Rural Web analysis framework confirms the theory from Van der Ploeg et al. (2008) about the differences that exist between the different actors in terms of their web. It also helps to explain their particularities, as well as to foresee the possible development trajectory. Through this framework it was possible to identify the pattern of interrelationships, interactions, exchanges and mutual externalities within the different groups involved in the Candelilla wax value chain. The Rural Web analysis provided relevant information about territory specific characteristics of the evaluated rural producers. The information provided includes insights of the local governance system, official and unofficial leadership structures and the type of incentives traditionally applied by the rural producers. This information is relevant for developing actors because it serves as baseline to delineate their strategies to encourage the participation of rural actors in specific economic activities and integrated development projects. Based on the information from the case study, for private firms to establish a solid relationship with rural producers, it is necessary to gain the trust from the leaders of the communities instead of simply limit their contact with regional rural leaders or authorities. Private firms should be open to negotiate the business conditions according to the needs and interests of the specific community. Also as observed in the case, the participation of development practitioners (governments, universities and NGOs) in some regions was relevant for the improvement of social cohesion and living conditions of the rural producers. However, in order to improve their relevance with the rural community members, they might need to strengthen

International Food and Agribusiness Management Review

Arato et al.

Volume 20, Issue 1, 2017

their bonds with community leaders in order to gain their trust and encourage their participation in future development projects. The discussion about the dialog and opportunities to increase understanding, in the results section of this paper, was elaborated upon based on the observations and comments received from different actors during the interview process. These might serve as a baseline for the actors involved to define improvement strategies (Block et al., 2008; Gregoratti, 2011; Messely et al., 2013). However, the suggested recommendations might only be useful once the interests of each actor are fulfilled or when a common trade-off between all parties is agreed upon. As addressed by Van der Ploeg et al. (2000), synergy is a strategic element in many rural development experiences. It creates cohesion between activities, not only at farm level but also between different regions and other rural activities. The present research focused on the analysis of a value chain which is based on the commercialization of natural goods by independent small and medium rural producers that are obtaining improvement opportunities and benefits from integrating their operations along the value chain in a sustainable way. Based on the analysis we can argue that the Rural Web framework is suitable for analyzing the socio-economic configuration of value chains whose early stages are based on small and medium size rural actors. This because it identifies the social relations that occur at different dimensions, which in turn could foster or hinder its success. Given the characteristics and complexity of each value chain in relation to its product, production process, distribution channels and markets, this work opens possible pathways for further research the applicability of the Rural Web framework for other types of value chains in different contexts.

References Anderson, A.R. and S.L. Jack. 2002. The articulation of social capital in entrepreneurial networks: a glue or a lubricant? Entrepreneurship and Regional Development 14: 193-210. Arato, M. 2016. Integration of private firms in rural development strategies: the case of rural communities from the Chihuahuan Desert in Mexico. PhD thesis, Ghent University, Ghent, Belgium. Arato, M., S. Speelman and G. Van Huylenbroeck. 2014. The contribution of non-timber forest products towards sustainable rural development: the case of Candelilla wax from the Chihuahuan Desert in Mexico. Natural Resources Forum 38: 141-153. Arnold, J.E.M. and M. Ruiz PĂŠrez. 1998. The Role of Non-Timber Forest Products in Conservation and Development, in: Incomes from the forest. Methods for the development and conservation of forest products for local communities. Edited by E. Wollenberg and A. Ingles CIFOR, Bogor, Indonesia, pp. 17-41. Barrett, C.B., D.R. Lee and J.G. McPeak. 2005. Institutional arrangements for rural poverty reduction and resource conservation. World Development 33: 193-197. Barrett, C.B., K. Brandon, C. Gibson and H. Gjertsen. 2001. Conserving tropical biodiversity amid weak institutions. BioScience 51: 497-502. Belcher, B., K. Schreckenberg. 2007. Commercialization of non-timber forest products: a reality check. Development Policy Review 25: 355-377. Belcher, B., M. RuĂz-PĂŠrez and R. Achdiawan. 2005. Global patterns and trends in the use and management of commercial NTFPs: implications for livelihoods and conservation. World Development 33: 1435-1452. Bellandi, M. 2001. Local development and embedded large firms. Entrepreneurship and Regional Development 13: 189-210. Bitzer, V., P. Glasbergen and B. Arts. 2013. Exploring the potential of intersectoral partnerships to improve the position of farmers in global agrifood chains: findings from the coffee sector in Peru. Agriculture and Human Values 30: 5-20. Blandon, J., S. Henson and J. Cranfield. 2009. Small-scale farmer participation in new agri-food supply chains: case of the supermarket supply chain for fruit and vegetables in Honduras. Journal of International Development 21: 971-984. Block, D.R., M. Thompson, J. Euken, T. Liquori, F. Fear and S. Baldwin. 2008. Engagement for transformation: value webs for local food system development. Agriculture and Human values 25: 379-388. International Food and Agribusiness Management Review

Arato et al.

Volume 20, Issue 1, 2017

Boettke, P., C. Coyne and P. Leeson. 2008. Institutional stickiness and the new development economics. American Journal of Economics and Sociology 67: 331-358. Bolwig, S., S. Ponte, A. Du Toit, L. Riisgaard and N. Halberg. 2010. Integrating poverty and environmental concerns into value-chain analysis: a conceptual framework. Development Policy Review 28: 173-194. Borgatti, S.P. and X. Li. 2009. On social network analysis in a supply chain context. Journal of Supply Chain Management 45: 5-22. Bowles, S. and H. Gintis. 2002. Social capital and community governance. Economic Journal 112: 419-436. CONABIO. 2009. Evaluación del estatus de euphorbia antisyphilitica en México dentro de los apéndices de la CITES. Decimoctava reunión del Comité de Flora. Buenos Aires (Argentina), 17-21 de marzo del 2009. Available at: http://tinyurl.com/zt2gpa3. CONAFOR. 2008. Catálogo de recursos maderables y no maderables. Clima Árido, tropical y templado. Available at: http://tinyurl.com/h3ke9go. CONAFOR. 2012. Empresas forestales y cadenas productivas. Available at: http://tinyurl.com/zg4n58t. Da Silva, C.A. and H.M. de Souza Filho. 2007. Guidelines for rapid appraisals of agrifood chain performance in developing countries. Rome: food and agriculture organization of the United Nations. Available at: http://tinyurl.com/jupoq83. Durand, G. and G. Van Huylenbroeck. 2003. Multifunctionality and rural development: a general framework. In: Multifuncional agriculture: a new paradigm for European agriculture and rural development, edited by G. Van Huylenbroeck and G. Durand. Ashgate Pub Limited, Farnham, UK. Durlauf, S.N. 2002. On the empirics of social capital. The Economic Journal 112: F459-F479. Earth Council. 1994. The earth summit-Eco 92: different visions. Earth Council and the Inter-American Institute for Cooperation on Agriculture, San Jose, Costa Rica. Ellis, F. 2000. Rural livelihoods and diversity in developing countries. Oxford University Press, Oxford, UK. Ellis, F. and S. Biggs. 2001. Evolving themes in rural development 1950-2000s. Development Policy Review 19: 437-448. Fasse, A., U. Grote and E. Winter. 2009. Value chain analysis methodologies in the context of environment and trade research (No. 429). Discussion papers//School of Economics and Management of the Hanover Leibniz University. Available at: http://tinyurl.com/hnt2c6e. Fisher, R.J. and R. Dechaineaux. 1998. A methodology for assessing and evaluating the social impacts of non-timber forest products projects. In: Incomes from the forest. Methods for the development and conservation of forest products for local communities, edited by E. Wollenberg and A. Ingles. CIFOR, Bogor, Indonesia, pp. 189-199. Fukuyama, F. 1999. The great disruption. Simon and Schuster, New York, NY, USA. Gray, B.J., S. Duncan, J. Kirkwood and S. Walton. 2014. Encouraging sustainable entrepreneurship in climate-threatened communities: a Samoan case study. Entrepreneurship and Regional Development 26: 401-430. Gregoratti, C. 2011. Global nuts and local mangoes: a critical reading of the UNDP growing sustainable business initiative in Kenya. Agriculture and human values 28: 369-383. Horlings, L.G. and T.K. Marsden. 2012. Exploring the ‘new rural paradigm’ in Europe: eco-economic strategies as a counterforce to the global competitiveness agenda. European Urban and Regional Studies 21: 4-20. Jarosz, L. 2000. Understanding agri-food networks as social relations. Agriculture and human values 17: 279-283. Kaplinsky, R. and M. Morris. 2002. Handbook for value chain research, IDRC. Available at: http://tinyurl. com/ha9vgnt. Marsden, T. 2010. Mobilizing the regional eco-economy: evolving webs of agri-food and rural development in the UK. Cambridge Journal of Regions, Economy and Society 3: 225-244. Marsden, T. and H. Renting. 2003. Understanding alternative food networks: exploring the role of short supply chains in rural development. Environment and Planning 35: 393-411.

International Food and Agribusiness Management Review

Arato et al.

Volume 20, Issue 1, 2017

Marshall, E., K. Schreckenberg, A.C. Newton. 2006. Commercialization of non-timber forest products. Factors influencing success. Lessons learned from Mexico and Bolivia and policy implications for decision-makers. UNEP World Conservation Monitoring Centre, Cambridge, UK. Available at: http://tinyurl.com/jku9g8y. Messely, L., E. Rogge and J. Dessein. 2013. Using the rural web in dialogue with regional stakeholders. Journal of Rural Studies 32: 400-410. Metzger, M.D., J. Ickis, F. Leguizamón and J. Flores. 2010. Inclusion of low income sectors in Latin American agribusiness. International Food and Agribusiness Management Review 13: 1-14. Mexican United States. 2015. The political constitution of the Mexican United States. Article 27. Available at: http://tinyurl.com/j6avslk. Multiceras. 2013. Candelilla wax exporter company. Interviews with representatives and information. Available at: http://tinyurl.com/hu2jla4. Nemes, G. 2005. Integrated rural development. The concept and its operation (No. 0506). Institute of Economics, Hungarian Academy of Sciences. Available at: http://tinyurl.com/gvfmwmt. Oostindie, H., and R. van Broekhuizen. 2008. The Dynamics of Novelty Production. In: Unfolding webs: the dynamics of regional rural development edited by J.D. Van der Ploeg and T.K. Marsden. Van Gorcum, Assen, the Netherlands. Peterson, H.C. 2013. Fundamental principles of managing multi-stakeholder engagement. International Food and Agribusiness Management Review 16: 11-22. Porter, M.E. 2011. Competitive advantage of nations: creating and sustaining superior performance. Simon and Schuster, New York, NY, USA. Ray, C. 2000. Further ideas about local rural development: trade, production and cultural capital. Centre for Rural Economy, Department of Agricultural Economics and Food Marketing, University of Newcastle upon Tyne, Newcastle, UK. Sampson, R., J. Morenoff and F. Earls. 1999. Beyond social capital: collective efficacy for children. American Sociological Review 64: 633-660. Schneider, E. 2009. Trade survey study on succulent Euphorbia species protected by CITES and used as cosmetic, food and medicine, with special focus on Candelilla wax. 18th meeting of the Plants Committee. Commissioned by Bundesamt für Naturschutz, CITES Scientific Authority, Germany. Available at: http://tinyurl.com/jp4tmem. Shucksmith, M. 2010. Disintegrated rural development? Neo-endogenous rural development, planning and place-shaping in diffused power contexts. Sociologia Ruralis 50: 1-14. Syampungani, S., P.W. Chirwa, F.K. Akinnifesi, G. Sileshi and O.C. Ajayi. 2009. The miombo woodlands at the cross roads: Potential threats, sustainable livelihoods, policy gaps and challenges. Natural Resources Forum 33: 150-159. Tallontire, A. 2007. CSR and regulation: towards a framework for understanding private standards initiatives in the agri-food chain. Third World Quarterly 28: 775-791. Tallontire, A., M. Opondo, V. Nelson and A. Martin 2011. Beyond the vertical? Using value chains and governance as a framework to analyse private standards initiatives in agri-food chains. Agriculture and Human Values 28: 427-441. Trienekens, J.H. 2011. Agricultural value chains in developing countries a framework for analysis. International Food and Agribusiness Management Review 14: 51-82. Uzzi, B. 1997. Social structure and competition in interfirm networks: the paradox of embeddedness. Administrative science quarterly 42: 35-67. Van der Ploeg, J.D and T.K. Marsden. 2008. Unfolding webs: the dynamics of regional rural development. Van Gorcum, Assen, the Netherlands. Van der Ploeg, J.D., H. Renting, G. Brunori, K. Knickel, J. Mannion, T. Marsden, K. De Roest, E. SevillaGuzmán and F. Ventura. 2000. Rural development: from practices and policies towards theory. Sociologia Ruralis 40: 391-408. Van der Ploeg, J.D., R. Van Broekhuizen, G. Brunori, R. Sonnino, K. Knickel, T. Tisenkopfs, and H. Oostindie. 2008. Unfolding webs: towards a framework for understanding regional rural development. Van Gorcum, Assen, the Netherlands, pp. 1-28. International Food and Agribusiness Management Review

Arato et al.

Volume 20, Issue 1, 2017

Van der Ploeg, J.D, P.J.M. Verschuren, F.P.M. Verhoeven and J.H.M. Pepels. 2006. Dealing with novelties: a grassland experiment reconsidered. Journal of Environmental Policy and Planning 8: 199-218. Ventura, F., G. Brunori, P. Milone and G. Berti. 2008. The rural web: a synthesis. In: Unfolding webs: the dynamics of regional rural development edited by J.D. Van der Ploeg and T.K. Marsden. Van Gorcum, Assen, the Netherlands. Ventura, F. and F. Milone. 2005. InnovativitĂ contadina e sviluppo rurale: unâ&#x20AC;&#x2122;analisi neo istitutionale del cambiamento in agricoltura in tre regioni del Sud Italia. Franco Angeli, Milano, Italy. Yin, R.K. 2012. Applications of case study research. 3rd edition. Sage Publications, Thousand Oaks, CA, USA.

International Food and Agribusiness Management Review

OPEN ACCESS International Food and Agribusiness Management Review Volume 20 Issue 1, 2017; DOI: 10.22434/IFAMR2016.0084 Received: 15 April 2016 / Accepted: 9 October 2016

The supply chain of Brazilian maize and soybeans: the effects of segregation on logistics and competitiveness RESEARCH ARTICLE Andréa L.R. de Oliveiraa and Augusto M. Alvim aProfessor,

bProfessor,

School of Agricultural Engineering, University of Campinas (UNICAMP), Cândido Rondon Avenue 501, Campinas, SP 13083-875, Brazil

Business School, Pontifical Catholic University of Rio Grande do Sul (PUCRS), Av. Ipiranga 6681 Partenon, Porto Alegre, RS 91410-330, Brazil

Abstract Despite the significant advances of Brazilian agriculture, transportation and storage costs still constitute the main barriers to the Brazilian agribusiness. The aim of this article is to analyze the effect of segregation of maize and soybeans in the Brazilian transport and storage logistics, especially genetically modified grains. In the context of the guidelines of the Cartagena Protocol on Biosafety (CPB) as well as of the competitiveness in the international market, we develop a spatial equilibrium model in the form of a mixed complementarity problem. The competitiveness of Brazilian maize and soybeans on the international market is compromised by the inefficient logistics and slow responses to the demands of the CPB. The contribution of the paper is to evaluate how regulatory issues of a segment, in this case biotechnology, may interfere with logistic infrastructure projects. Keywords: agricultural logistics, partial equilibrium models, international agreements, biotechnology JEL code: Q13, Q02, Q17 Corresponding author: augusto.alvim@pucrs.br

De Oliveira and Alvim

Volume 20, Issue 1, 2017

1. Introduction Brazilian agribusiness stands out globally for its competitiveness in exporting products such as sugar, coffee, orange juice, ethanol, soybean and derivatives, and beef, pork and chicken meat (FAO, 2014; USDA, 2014). This performance is due mainly to three measures: the management of natural resources, the use of genetic engineering in the development of new varieties, and the adoption of new management practices. The continued success of Brazilian agriculture depends on successive advances in new technologies that ensure productivity gains and/or added value in the final product. The intensification of the production is associated with a high use of fertilizers and pesticides, with significant increases in labor and land productivity. Genetic engineering, notably the creation of genetically modified organisms (GMOs), and mechanization with intense automation and precision agriculture represent some of the technologies that have significantly increased agricultural productivity (Buainain, 2014). However, despite the gains recorded and potential use of biotechnology in sectors such as agriculture and medicine, this has been one of the most controversial issues of our time. This is due, in large part, to the disagreement among stakeholders concerning the actual or potential benefits and risks of products from agricultural biotechnology. Some social groups recognize the benefits of GMOs, such as the decreased use of pesticides, smaller agricultural production costs, increased yield per hectare, and lower food prices. However, many others focus on the potential negative effects of GMOs to the well-being of the population, including threat to human health and environmental damage or loss of biodiversity (Oliveira et al., 2012). Regardless of the opinion on GMOs, two points deserve emphasis. First, stricter rules regarding GMOs have economic implications, particularly in developing countries. The proliferation of biosafety systems and GMO authorization rules related to labeling, identity preservation, segregation, and traceability may complicate the international trade of genetically modified agricultural products and negatively affect the trade of agricultural commodities, especially in Brazil. Second, segregation affects logistics and increase storage and transportation costs (Schlecht et al., 2004). The full segregation for maize and soybeans requires more compartments in the storage units or lower-capacity silos that allow segregated storage. The theoretical design of GMO regulations follows the Convention on Biological Diversity through the Cartagena Protocol on Biosafety (CPB) and the World Trade Organization, which are, in turn, based on the standards and guidelines of the Codex Alimentarius to address the issues on food security. The process of setting up consensus-approved multilateral protocols is complex. There is little room for collective action in the context of the CPB and specific provisions about GMOs as advancing from thesis towards practice often involves deadlocks.1 The CPB is expected to have more intense effects on the soybean and maize markets. Both are products with a significant share in global agricultural production. The economic impact will depend on the costs of the resources needed to comply with the legal requirements of the CPB. The aim of this paper is to analyze â&#x20AC;&#x201C; in the context of the CPB â&#x20AC;&#x201C; how the segregation of maize and soybeans affects the logistics of transport and storage in Brazil and impacts the countryâ&#x20AC;&#x2122;s international competitiveness. Such analysis demands appropriate analytical tools and scenario simulations. The findings may guide the implementation of more effective policies and support new investments in logistics.

1 Common

sources of disagreement are labeling, transport, liability, and redress.

International Food and Agribusiness Management Review

De Oliveira and Alvim

Volume 20, Issue 1, 2017

2. Grain segregation and agroindustrial logistics Over 369 biotechnology events2 and 27 biotech crops have passed the regulatory barrier in several countries as of June 2015, and many of them have been marketed during the last fifteen years (ISAAA, 2015). Farmers around the world adopted biotech crops swiftly. The planted area of biotech crops increased approximately 62-fold since 1996 (James, 2013). Brazil had remained the second largest producer of biotech crops in the world in 2013. The planted area of transgenic crops increased by 3.7 million hectares in one year (the largest absolute increase observed in any country in the world) to 40.3 million hectares in 2013 (James, 2013). Brazil implemented a faster approval system and approved 18 transgenics events (GM events) between 2010 and 2013. The commercialization of the first soybean variety resistant to insects and tolerant to herbicides was approved in 2012. Notably, Brazilian Agricultural Research Corporation (EMBRAPA) demonstrated its impressive technical capacity to develop and release a new cutting-edge GM crop by developing a virus-resistant locally produced bean entirely with its own resources and receiving approval for marketing it (James, 2013). The Cartagena Protocol on Biosafety (CPB) governs the regulation on GMOs. Article 18 of the CPB establishes the rules for handling, transportation, packaging, and identification of all loads that contain or may contain GMOs. The analysis in this paper is restricted to loads of genetically modified organisms intended for direct use as food, feed, or for processing (GMOs-FFPs) (Mackenzie et al., 2003). The mandatory implementation of processes that lead to an increase in fixed costs, with no direct connection to the fulfillment of the objectives of the Cartagena Protocol (CPB) – especially through Article 18.2 – should be viewed as a new component in the process of creating technical barriers to trade, with negative effects on agricultural producers in exporting countries and on consumers in importing countries. Based on this principle, the CPB establishes, in Article 18, the requirements and necessary steps with regard to handling, transport, packaging and identification of all loads that contain or may contain living modified organisms (LMOs). The purpose of this analysis is restricted to loads of LMOs-FFPs, whose requirements are set out in paragraph 2.a of Article 18 (Mackenzie et al. 2003): 2. Each Party will take measures to require that documentation accompanying: (a) living modified organisms intended for direct use as food or feed, or for processing, clearly identifies that these ‘may contain’ LMOs and are not intended for intentional introduction into the environment, as well as a contact point for further information. The Conference of the Parties serving as the meeting of the Parties to this Protocol will take a decision on the detailed requirements for this purpose, including specification on its identity and any unique identification, no later than two years after the entry into force of this Protocol (Brasil, 2006). Although in its original text the Protocol uses the expression ‘may contain’, most importers of agricultural products requires the load to be identified with the use of the term ‘contain’. The analytical framework aims to assess the intervention of CPB in Brazilian exports. It borrows from Gruère and Rosegrant (2008) who rated countries to assess the impact of Article 18.2.a. of the CPB. Gruère and Rosegrant (2008) divided the countries into four groups according to the countries’ affiliation to the CPB and the production of GMOs:

2 These

biotechnological advances include creating genetically modified products (GM event) or transgenic products.

International Food and Agribusiness Management Review

De Oliveira and Alvim

■■ ■■ ■■ ■■

Volume 20, Issue 1, 2017

Group 1: countries that produce GMOs, but are not parties to the CPB (for example, Argentina, United States, Canada). Group 2: countries that do not produce GMOs, but are parties to the CPB (for example, Japan, UK, Peru). Group 3: countries that produce GMOs and are parties to the CPB (for example, Brazil, China, South Africa). Group 4: countries that do not produce GMOs and are not parties to the CPB (for example, Russia, Israel, Chile).

Important observations may be inferred from a simple depiction of how the imposition of the CPB affects trade flows (Figure 1). First, the trade flows in countries in Group 3 (parties to the Protocol and producers of GMOs) are the most affected. Those countries would have to control the imports from all GMO-producing countries and exports to the other parties to the CPB as well as the exports to all non-member countries. Second, Group 2 countries (parties to the Protocol that do not produce GMOs) would have to control only the imports from countries in Group 1. The imposition of the CPB impacts Brazil (Group 3) more than the United States and Argentina (both Group 1), the main competitors of Brazil in the soybean market (Figure 1). The soybean export competitiveness of Brazil is thus negatively affected. All flows of the Brazilian exports of grain are similarly affected. The main importers of Brazilian maize and soybeans belong to Groups 2 and 3 (parties to the CPB), hence, Brazil would have to carry out costly qualitative and, in many cases, quantitative tests to identify GMO shipments. These costs may result in loss of trade. According to Gruère and Rosegrant (2008), the identification requirements in Article 18.2.a. of the Cartagena Protocol impose additional marketing costs in maize and soybeans exports and affect the exporting countries’ competitiveness. Oliveira et al. (2012) gauged the implementation of the CPB as for the Brazilian soybean to fit the term ‘contains GMO.’ The study indicated that, considering the cost of tests to identify two transgenic events, segregated logistics, and reduction of international trade, the total competitiveness loss for Brazil in the

Group 1 GM and non-CPB

Group 3 GM and CPB

Group 4 non-GM and non-CPB

Group 2 non-GM and CPB regular flow of GM and non-GM affected flow: export and import affected flow: export and import

Figure 1. Trade flows affected by the implementation of the Cartagena Protocol on Biosafety (adapted from Gruère and Rosegrant, 2008). International Food and Agribusiness Management Review

De Oliveira and Alvim

Volume 20, Issue 1, 2017

international market reached US$ 1.57 billion. This amount represents 13.8% of the foreign currency generated by Brazilian soybean exports in 2009. International trade loss may be even more significant for maize, considering that the number of transgenic events requiring evaluation is superior to soybean (six on average). Further, there is more spatial fragmentation in maize production than soybean production. This makes establishing routes very difficult and complicates the calculation of the impact of the full segregation of conventional and transgenic loads. The movement of homogeneous and standardized products proved to be an important strategy to ensure economies of scale and facilitate logistics. However, the demand for differentiated grains has grown significantly. The exports of Brazilian agricultural commodities are at a disadvantage compared with commodities produced in other countries due to the high costs of transportation via inadequate roads and costly and inefficient port services. In addition to the issues with the transportation system, the storage infrastructure in Brazil adds to the logistics. Despite the growing investment in storage in Brazil, it has not accompanied the growth of the agricultural sector. After a small annual growth of 2.1%, the static capacity of warehouses was 145.6 million tons, but did not accommodate the grain production of 186.9 million tons for the 2012/2013 harvest (an annual increase of 12.5%) (CONAB, 2014b). The storage deficit thus amounted to 22.1% (Figure 2). Adding the challenges of CPB implementation to the critical shortage of warehouses for conventional production exacerbates the already inherent limitations of proper storage for segregated beans. Improvements in transportation should accompany the expansion of farm areas originating a new, larger spatial arrangement for the production sectors. The transportation sector thus constitutes a logistical bottleneck. Harnessing the potential of grain production, however, will be possible only through an efficient road system, integrated intermodal transport corridors, and addressing the storage deficit, especially for the segregated cargo.

3. Methodology In order to quantify the potential impact of the costs of implementing the CPB for Brazil, with a focus on the organization of the Brazilian logistics of transport and storage, we used a partial equilibrium model formulated as a mixed complementarity problem (MCP). The use of MCP has been proposed by Thore (1991), Rutherford (1995) and Bishop et al. (2001), and it has already been used by Oliveira et al. (2012) and Alvim and Waquil (2004). Warehouse capacity 186,9

/1 1 20 11 /1 2 20 12 /1 3

20 0

2/ 20 0

1/ 20 0

0/ 20 0

145,6

(million tons)

Grain production 200 180 160 140 120 100 80 60 40 20 0

Figure 2. Evolution of the static storage capacity and grain production in Brazil, 2000/2001-2012/2013 harvest (adapted from CONAB, 2014a) International Food and Agribusiness Management Review

De Oliveira and Alvim

Volume 20, Issue 1, 2017

The MCP has the advantage of allowing the easier incorporation of rates, Oliveira et al. (2012) analyzed the main effects of segregation of soybean market on the Brazilian logistics systems, included the cost of segregation like additional rate in the MCP. The model allows the simulation of different scenarios, from the most restrictive to the more relaxed. Alvim and Waquil (2004) has used the MCP to examine possible changes in the rice market through the implementation of some trade policies. The main changes occur in production, consumption, prices, trade flows and consumer and producer surplus when are simulated changes in tariff barriers and subsidies. The partial equilibrium models elect a sector or product under consideration and examine the effects of a (exogenous) variation of the relative price on the balance of the industry, assuming that the allocation for the rest of the economy remains unchanged (Ahumada and Villalobos, 2009). We adopt the partial equilibrium approach to analyze the impact of CPB in the Brazilian soybean and maize market. The method allows for a detailed evaluation of the effects of implementing the CPB in the Brazilian commercial flows. Another advantage of this method is an easier incorporation of tariffs, tariff quotas, and grants. A complementarity problem consists of a system of simultaneous equations (linear or nonlinear inequalities) derived from the functions of supply and demand. The MCP is equivalent to the Kuhn-Tucker conditions, which are necessary and sufficient to maximize the net social payoff function3 (NSP). Maximizing the NSP function implies achieving equilibrium in all markets and in all regions.4 It is easier to incorporate rates, quotas, and grants in the MCP (Bishop et al., 2001). The MCP for analyzing the Brazilian grain market is as follows: J

0 ≤ φi ┴ ∑ xij + ∑ xik ≤ zi (1),5 j

0 ≤ λj ┴ yj ≤ ∑ xij (2) i

0 ≤ µk ┴ yk ≤ ∑ xik

(3)

0 ≤ xij ┴ pi + tij ≥ pj ∀i,j (4) 0 ≤ xik ┴ pi + tik ≥ pk ∀i,k (5) Where i denotes supply regions (i = 1,…,n), j denotes domestic demand regions (j = 1,…,m), k denotes international demand regions (k = 1,…,g), and r denotes transport routes (r = 1,…,h). pi stands for observed supply prices, pj for observed domestic demand prices, and pk for observed international demand prices. zi represents the quantity supplied by region i, yj the quantity consumed in domestic market j, yk the quantity consumed in international market k, and xi,j total trade from region i to region j. The parameters of the model are transportation cost ti,j; shadow price associated with the supply region i, φi; shadow price associated with the demand region j, λj; and shadow price associated with the demand region k, µk. 3 Samuelson’s

(1952) formulation shows that maximizing the NSP function, given by the sum of the surplus of producers and consumers minus shipping cost and subject to regional balance equations, generates a framework of optimality conditions. Samuelson warned about problems associated with the use of his model to make inferences about social welfare. Hence, the expression ‘net social payoff’, which excludes a reference to social welfare (Samuelson, 1952). 4 The global maximum is the solution to a problem of nonlinear programming with a differentiable and concave objective function and linear differentiable and convex constraints, since the optimal point satisfies the Kuhn-Tucker conditions (Takayama and Judge, 1971). 5 In the MCP, elasticity coefficients are included in the restrictions (1), (2), and (3) such that the quantities produced and consumed are replaced by the following expressions: zi = ai × φbi i

j yj = cj × λ-d j

yk = ek × µk-fk

International Food and Agribusiness Management Review

De Oliveira and Alvim

Volume 20, Issue 1, 2017

The ‘┴’ symbol conveys that at least one of the adjacent inequalities must be satisfied as a strict equality. This is nothing more than a formality of the complementarity in the Kuhn-Tucker conditions. Equations (4) and (5) thus facilitate the inclusion of the ad valorem rate or tariff entailed by the cost of the test to identify transgenic events. The inclusion of the ad valorem rate follows Bishop et al. (2001). A parameter taxik, representing an ad valorem rate or tariff, may be incorporated in the model in Equation (5), considering the zero net condition. This is because, in this study, the rate has implications only for flows earmarked for the international market. Modifying the zero-net condition results in: (pi + tik) (1 + taxik) ≥ pk ∀i,k

(6)

The rate taxik represents the cost of tests for the identification and quantification of GMO events, plus the cost of segregated storage on the international market flows, as the CPB imposes measures on trans-boundary movements. In equilibrium, if there is a trade flow between producing regions and international demand, the price of the product in the region of supply plus the cost of transport, after the imposition of the GMO tests and segregation, should be equal to the price of the international demand. Otherwise, in the absence of commercial flows, the price in the region of international demand is lower than the price in the region of supply plus transportation costs and tests. The starting point for our model was to identify and select regions of supply and demand for soybean and maize. The evolution over the last few years of several variables, such as production, average yield, cultivated area, exports, industrial processing, and consumption for pig and poultry production, was analyzed to capture the dynamics in Brazil’s prominent agricultural regions with great potential for expansion. Regions of excess of supply are regions where soybean or maize production is greater than the amount processed. Otherwise, they were regions of excess of demand. Regional heterogeneity in the states of Mato Grosso and Paraná implies different trade flows and the use of different transportation routes; hence, different production and processing micro regions6 were identified in these states. Source of data The model employs data from 2011. Production data are from the Brazilian Institute of Geography and Statistics and the United States Department of Agriculture (USDA). The Brazilian Association of Chick Producers and the Brazilian Association of Pork Producers and Exporters were the sources for consumption data. The consulting agency Safras and Mercado (2011) and USDA (2011) were the source for soybean and maize prices for the domestic and international markets, respectively. Studies by Fuller et al. (2001, 2003) and FAPRI (2011) were the source of data on price elasticities of supply and demand. The freight from the road, rail, waterway, and maritime transportation modes come from the freight information system, Sifreca (2011). We conducted personal interviews with semi-structured questionnaires with key industry players (trading companies, shipping companies and certifiers). This exploratory and qualitative research allowed us to understand the operational aspects of segregation of GMO grains and to obtain the segregated storage costs and costs of tests to identify transgenic events. In addition, some trading companies authorized visits to their facilities, which allowed the observation and viewing of all stages of segregation operations, including the boarding of ships. The interviews were conducted with key players in the grain chain. This methodological approach known as ‘rapid assessment’ or ‘quick appraisal,’ as Dunn (1994) in which use data from secondary sources together 6 Microregions

consist of a cluster of cities with similar agricultural, industrial, and economic characteristics.

International Food and Agribusiness Management Review

De Oliveira and Alvim

Volume 20, Issue 1, 2017

with non-random samples and semi structured interviews with the key players can be applied in research that is necessary to obtain data and/or more detailed information to understand the dynamics of the sector. The interview guide was directed to three different groups. In the first group, the agricultural trading companies, agricultural cooperatives and rural producers association have been interviewed. The second group were the railway undertakings, including trading companies who own businesses and/or owns part of logistics operations, such as the Caramuru and Amaggi that control waterway companies like Torque S.A. and Hermasa Amazon Navigation SA, respectively. The third group were the main laboratories and certifiers. The interviews were conducted during the second half of 2009. Respondents were representatives of ADM, Amaggi, Bunge, Caramuru, Cargill, Brazilian Association of Non-Genetically Modified Grain Producers (EXTEND), Cocamar, Cooperative Castrolanda. As for logistics operators were America Latina Logistica (ALL), Companhia Vale do Rio Doce (Vale), S.A. Torque and Hermasa Amazon Navigation S.A. Finally, the laboratories and certifiers: CertID Brazil, Eurofins and SGS Brazil. The field research helped determine the cost of tests for the identification of transgenic events (GM events) as well as the sampling pattern. There are two methods to analyze GMOs: DNA analysis and protein analysis. The DNA analysis uses a quantitative or qualitative technique named polymerase chain reaction (PCR). The protein analysis employs a simple enzyme-linked immunosorbent assay (the dipstick test) and detects only one event at a time. The unit cost was US$ 3 for the dipstick test and US$ 300 for the PCR. Two samples are taken from every 40 tons and requiring two dipstick tests with a total cost of US$ 6. The quantitative real-time PCR for maize cost US$ 1,050 per sample and six GM events; three analyses7 for every 3,000 tons amount to a total of US$ 3,150. The PCR cost for soybean was US$ 900, with only one event tested. Major companies exporting non-GM maize and non-GM soybean were sources for segregated storage costs. Costs in trans-shipment warehouses were approximately 13 US$/ton and storage at ports of export were, on average, 10 US$/ton. The costs of tests and storage were estimated for grain labeled as ‘contains GM.’ Considering the descriptor ‘may contain GM’ would not significantly impact marketing costs or the logistics structure (Borges et al., 2009; Gruère and Rosegrant, 2008; Huang et al., 2008; Kalaitzandonakes, 2004; Simões, 2008). Table 1 describes the supply and demand regions and Table 2 shows the logistical routes considered and analyzed in the research. In Table 1 is shown the main producer and consumer regions in Brazil and also the main importers of corn and soybeans. In turn Table 2 describes the Brazilian logistical routes obtained from interviews conducted with key players in the grain chain and representing the major transport routes used in Brazil. Alternative scenarios We simulated two different scenarios. Scenario 1 is the control, in which there are no expenses with GMO tests and segregated storage; trade flows were based only on transportation costs (i.e. without the CPB rules concerning the term ‘contain’ in place). Scenario 2 represents a framework of full segregation of grain that ‘contains GM.’ The PCR test was considered when boarding on the ship. The number of dipstick tests varied according to the transport route considered. Each change of transportation mode requires trans-shipment operations to prevent the mixing of cargo and an additional dipstick test. The process also considers segregated storage. Consequently, the 7 One

PCR is performed when boarding, one at the port of export and one on the ship.

International Food and Agribusiness Management Review

De Oliveira and Alvim

Volume 20, Issue 1, 2017

Table 1. Description of the supply and demand regions considered in the model. Regions Description

Classification

PR-N PR-W PR-SE RS MG SC SP GO MS MT-N MT-W MT-SE MT-NE Europe

supply region supply region domestic demand region supply region supply region domestic demand region domestic demand region supply region supply region supply region supply region domestic demand region supply region international demand

China Japan Taiwan Iran

north of Paraná State west of Paraná State southeast of Paraná State Rio Grande do Sul State Minas Gerais State Santa Catarina State São Paulo State Goias State Mato Grosso do Sul State north of Mato Grosso do Sul State west of Mato Grosso do Sul State southeast of Mato Grosso do Sul State northeast of Mato Grosso do Sul State European Union (EU 27): Germany, Austria, Belgium, Bulgaria, Cyprus, Denmark, Slovakia, Slovenia, Spain, Estonia, Finland, France, Greece, Hungary, Ireland, Italy, Latvia, Lithuania, Luxembourg, Malta, the Netherlands, Poland, Portugal, United Kingdom, Czech Republic, Romania and Sweden.

international demand international demand international demand international demand

Table 2. Description of the logistical routes considered in the model. Route

R1 R2 R3 R4 R5 R6 R7 R8 R9 R10 R11 R12 R13 R14 R15 1

Description Destination1

Transport modals

SP demand PR-SE demand MT-SE demand port of Santos port of Santos port of Santos port of Santos port of Santos port of Paranaguá port of Paranaguá port of Rio Grande port of Rio Grande port of Rio Grande port of Vitória port of Santarem

road road road road road and rail road and rail road and rail road-hydro-rail road road and rail road road and rail road and hydro road and rail road and hydro

Transshipment points

rail terminal of Alto Araguaia rail terminal of Campo Grande rail terminal of Goiânia hydroport of São Simão and rail terminal of Pederneiras rail terminal of Londrina rail terminal of Cruz Alta hydroport of Estrela rail terminal of Araguari hydroport of Porto Velho

SP = São Paulo, PR = Paraná, MT = Mato Grosso.

International Food and Agribusiness Management Review

De Oliveira and Alvim

Volume 20, Issue 1, 2017

ad valorem rate calculated was 60% in intermodal flows and 55% in unimodal flows. These data are also the result of interviews conducted with key players in the grain chain. The study employs the General Algebraic Modeling System software (GAMS Development Corporation, Washington, WA, USA) to process the information for the movement of maize and soybean in Brazil within the MCP framework (Brooke et al., 1995).

4. Results and discussion Mathematical programming models must be validated by checking the consistency of the results of the problem. In Waquil and Cox (1995), the validation presupposes an adaptation of the coefficients and the structure of the model. How well the model’s solution approaches the real situation can validate the model. Many spatial equilibrium models generate results different from the actual data (Thompson, 1981), without invalidating the model. Tables 3 and 4 present the levels of supply and demand estimated by the model (scenarios 1 and 2) for soybean and maize, respectively, as well as the observed quantities in 2011 (observed data). Scenario 1 corresponds to the control group with no expenses regarding GMO tests and segregated storage; trade flows were based only on transportation costs. This scenario represents business transactions without the imposition of the CPB. Table 3. Volumes of supply, domestic demand and international demand of soybean.1 Regions

Supply Total Mato Grosso (MT) North MT Northeast MT West MT Total Paraná (PR) West PR North PR Rio Grande do Sul Goiás Minas Gerais Mato Grosso do Sul Total supply Domestic demand (D) São Paulo Southeastern PR Southeastern PR Subtotal International Demand (E) China EU-27 Japan Subtotal Total demand (D+E) 1

Observed data (A) (thousands tons)

Scenario 1 (B) (thousands tons)

Scenario 2 (C) (thousands tons)

Variation B to C (%)

13,131.2 7,071.4 3,792.0 2,267.9 6,525.5 3,671.1 2,854.4 4,853.8 2,820.4 1,282.9 1,065.9 29,679.6

13,489.0 7,222.4 3,934.1 2,332.5 6,692.4 3,753.4 2,939.0 4,924.7 2,931.1 1,307.9 1,073.4 30,418.4

12,930.9 6,926.5 3,762.7 2,241.8 6,411.9 3,592.6 2,819.3 4,801.7 2,789.9 1,257.2 1,016.8 29,208.4

-4.14 -4.10 -4.36 -3.89 -4.19 -4.28 -4.07 -2.50 -4.82 -3.87 -5.27 -3.98

2,028.9 1,225.0 748.5 4,002.4

2,023.4 1,221.4 745.0 3,989.7

2,046.0 1,234.7 754.6 4,035.4

1.12 1.09 1.30 1.15

17,000.0 10,000.0 700.0 27,700.0 31,702.4

16,232.7 9,558.9 637.2 26,428.7 30,418.4

15,522.0 9,068.0 583.0 25,173.0 29,208.4

-4.38 -5.14 -8.50 -4.75 -3.98

The tabel contains model estimates (scenarios 1 and 2) and observed data from 2011.

International Food and Agribusiness Management Review

De Oliveira and Alvim

Volume 20, Issue 1, 2017

Table 4. Volumes of supply, domestic demand and international demand of maize. Regions

Supply Total Mato Grosso (MT) North MT Southeast MT Goiรกs Mato Grosso do Sul Minas Gerais Total Paranรก (PR) North PR West PR Total supply Domestic demand (D) Santa Catarina Rio Grande do Sul Sรฃo Paulo Subtotal International demand (E) Iran Japan Taiwan Subtotal Total demand (D+E) 1

Observed Data (A) (thousands tons)

Scenario 1 (B) (thousands tons)

Scenario 2 (C) (thousands tons)

Variation B to C (%)

6,610.1 4,957.6 1,652.5 4,004.0 2,747.7 3,386.0 5,367.8 3,489.1 1,878.7 22,115.5

5,084.3 3,827.0 1,257.3 3,003.3 1,962.5 2,704.3 4,293.4 2,841.0 1,452.4 17,047.8

4,826.2 3,627.0 1,199.2 2,875.9 1,848.9 2,610.6 4,121.0 2,677.8 1,443.2 16,282.6

-5.08 -5.22 -4.62 -4.24 -5.79 -3.47 -4.02 -5.74 -0.64 -4.49

3,028.2 124.7 270.7 3,423.6

3,179.4 130.8 288.6 3,598.7

3,210.7 130.9 289.1 3,630.8

0.99 0.13 0.19 0.89

5,000.0 4,000.0 4,000.0 13,000.0 16,423.6

5,329.7 4,016.3 4,103.1 13,449.1 17,047.8

4,796.9 3,983.3 3,871.6 12,651.9 16,282.6

-10.00 -0.82 -5.64 -5.93 -4.49

The table contains model estimates (scenarios 1 and 2) and observed data from 2011.

In scenario 2, the system for the identification and quantification decreased soybean production by 3.98%. International flows were the most affected, with losses amounting to 1.25 million tons. The exports to Japan and Europe, the main importers of non-GM soybean, fell by 8.50 and 5.14%, respectively. Scenario 2 (full segregation) for soybean gives evidence of the loss of competitiveness of the Brazilian soybean in a system of segregation. Adjusting the parameters of the model allows for simulating changes in production performance and consumption in the regions under analysis as an international agreement is simulated. Due to the expenses with tests and storage (US$ 1.1 billion) and the reduction of international trade (US$ 482.8 million), monetary losses reached US$ 1.57 billion. This amount represents 10% of the foreign currency generated by exports of Brazilian soybean in 2011 (US$ 16.33 billion according to the Ministry of Development, Industry and Foreign Trade (MDIC)). In scenario 2 for maize, the system for the identification and quantification of transgenic maize events decreased trade by 4.49%. The greatest effect was on international flows and losses were 765,000 tons. The exports to Iran and Taiwan, the main partners of Brazil, fell by 10% and 5.64%, respectively. The most effect was on regions far from ports of export, as any increase in the costs of logistics is felt more strongly. The most affected states were Mato Grosso do Sul (central-western region of Brazil), 5.79%, followed by the north of Paranรก (southern region of Brazil), 5.74%, and the north of Mato Grosso (centralwestern region of Brazil), 5.22%. International Food and Agribusiness Management Review

De Oliveira and Alvim

Volume 20, Issue 1, 2017

Scenario 2 gives evidence of the loss of competitiveness of the Brazilian maize exports and of the regional impact. The Brazilian production system requires a greater number of transfers, given the long distances to ports of export. Brazil’s main competitors – the United States and Argentina – have greater logistical efficiency, thus, the reduction of the Brazilian competitiveness becomes eminent. Due to the expenses with tests and storage (US$ 506 million) and the decrease in international trade (US$ 212 million), monetary losses regarding maize reached US$ 718 million. This amount represents 26% of the foreign currency generated by Brazilian maize exports in 2011 (US$ 2.72 billion according to MDIC). The simulation of scenario 2 suggests that the CPB measures may have very different effects in each of the major regions of maize production and exports in Brazil. The losses in this scenario ranged from 0.64 to 5.79%. These differences arise from the different storage and transport infrastructures. Tables 5 and 6 show the trade flows and the logistics routes used for moving soybean in scenarios 1 and 2, respectively. In scenario 1, the Chinese and Japanese markets were the targets of production for this region, using intermodal transport through the port of Santarém. Soybean was transported by truck to the river port of the city Porto Velho, and from there travelled by waterway to the port of Santarém (Figure 3). In scenario 2, there were changes in the transacted volumes and target markets. In scenario 1, the exports were made via intermodal options, responsible for 86.9% of movements (26.4 million tons). In scenario 2, only 15% of the soybean for the international market was carried by intermodal options (approximately 3.7 million tons). Only exports from the western region of Mato Grosso used intermodality as a competitive option. Due to the implementation of segregation measures, more than 77% of intermodal routes ceased to be competitive (increased cost). Priority for the road mode was overwhelming and the costs of implementing the CPB had a greater impact on intermodal routes because of the greater number of tests Table 5. Soybean: trade flow by transportation route, scenario 1.1,2 Supply

Demand

Route3 R1

PR-W PR-SE MT-N MT-SE GO SP RS EU-27 PR-N China PR-W EU-27 MT-N China MT-NE EU-27 MT-NE China MT-W China MT-W Japan MS China GO China MG China Total = 30,418.4

R2 1,221.4

2,023.4

R10

R13

R14

R15

745.0

2,939.0 6,477.5 2,102.2 230.3

907.7 2,023.4 1,221.4 745.0

8,810.0 907.7

4,924.7 2,532.0

1,073.4

3,296.9 637.2

1,307.9 2,939.0 3,605.4 4,924.7 1,307.9 3,934.1

Road route (unimodal): R1, R2, R3, R9; intermodal route: R5, R8, R10, R13, R14, R15. PR = Paraná; MT = Mato Grosso; GO = Goiás; SP = São Paulo; RS = Rio Grande do Sul; MS = Mato Grosso do Sul; MG = Minas Gerais. 3 Values in thousands of tons. 2

International Food and Agribusiness Management Review

De Oliveira and Alvim

Volume 20, Issue 1, 2017

Table 6. Soybean: trade flow by transportation route, scenario 2.1,2 Supply

Demand

Route3 R1

PR-W SE MT-N MT-SE GO SP RS EU-27 PR-N China PR-W China MT-N China MT-N UE-27 MT-N Japan MT-NE China MT-W China MS UE-27 MG China GO EU-27 Total = 29,208.4

R2 1,234.7

2,046.0

1,234.7

R11

R15

754.6

3,083.2 2,505.6 583.0 2,241.8

2,046.0

754.6

1,257.2 743.9 10,414.7

2,819.3 2,357.8

4,801.7

3,762.7

1,016.8

6,193.9

4,801.7

3,762.7

Road Route (unimodal): R1, R2, R3, R4, R9, R11; intermodal route: R15. 2 PR = Paraná, SE = Sergipe, MT = Mato Grosso, GO = Goiás, SP = São Paulo, RS = Rio Grande do Sul, MS = Mato Grosso do Sul, MG = Minas Gerais. 3 Values in thousands of tons.

Supply GO MG Domestic demand

International demand

MT-N

China

MT-SE

MT-NE

EU-27

PR-SE

MT-W

Japan

PR-N PR-W RS

Intermodal route Road route

Figure 3. Soybean: trade flow by transportation route, scenario 1. SP = São Paulo; MT = Mato Grosso; PR = Paraná; GO = Goiás; MG = Minas Gerais; MS = Mato Grosso do Sul; RS = Rio Grande do Sul.

International Food and Agribusiness Management Review

De Oliveira and Alvim

Volume 20, Issue 1, 2017

and increasing demand for segregated storage. The cost of soybean transport thus increased in comparison with unimodal, railroad-only routes. The inefficiency of the Brazilian logistics is represented by high transportation costs. Moreover, the limitations of intermodality are evident by the high cost of transshipment. The result is that segregation make the most inefficient Brazilian logistics, especially intermodal flows that are the most affected. In scenario 1 for maize, a portion of the maize production in Minas Gerais (MG) was destined for the domestic market, supplying Santa Catarina in the southern region of Brazil and using only road transport (route R2). Another portion of the production went to Iran through the port of Vitรณria. The road and railway routes were used (intermodal route) for this flow. Maize was transported by trucks up to the rail terminal located in the city of Araguari (MG), and from there it was transported by rail up to the port of Vitรณria (route R14) (Table 7). An increase in logistics costs due to the CPB rules modified movement of maize in scenario 2 (Table 8).The region of Minas Gerais began providing a greater volume to local markets and started to export maize to Taiwan. In addition, the route for the international market changed (the intermodal route (R14) used before in scenario 1 was no longer competitive). The export was made to the port of Santos via road (route R4). In scenario 1, the exports of maize were through intermodal options, responsible for 100% of the movements (13.45 million tons). In scenario 2, only 29% of the maize for the international market was carried by intermodal options (approximately 3.6 million tons). Only exports from the northern region of Mato Grosso used intermodality as a competitive option. The increased transportation cost was compared to unimodal railroad-only routes. The logistics of transport and storage are affected by the requirements of the CPB. Therefore, the more rigid the identification process, the greater the impact on exports. Consequently, the competitiveness of Brazilian maize on the international market suffers due to inefficient logistics, that are unresponsive to the demands of the CPB. Table 7. Maize: trade flows by transportation route, scenario 1.1,2 Supply

Demand

Route3 R1

PR-W SC PR-W SP PR-W RS MS SC MG SC PR-N Iran PR-W Japan MT-N Taiwan MT-SE Taiwan MT-SE Japan MG Iran GO Japan Total = 17,047.7

288.6

R10

R15

1,001.3

1,962.5 215.7

130.8

2,841.0 31.8 276.1 981.2

288.6

R14

3,179.4

130.8

3,003.3 4,260.6

3,827.0

2,488.7 2,872.8

2,488.7

3,827.0

Road Route (unimodal): R1, R2, R3; intermodal route: R8, R10, R14, R15. 2 PR = Paranรก; SC = Santa Catarina; RS = Rio Grande do Sul; MS = Mato Grosso do Sul; MG = Minas Gerais; MT = Mato Grosso; GO = Goiรกs. 3 Values in thousands of tons. International Food and Agribusiness Management Review

De Oliveira and Alvim

Volume 20, Issue 1, 2017

Table 8. Maize: trade flows by transportation route, scenario 2.1,2, Supply

Demand

Route3 R1

PR-W RS MS SP MS SC MG SC PR-N Iran PR-W Iran MT-N Japan MT-SE Iran MT-SE Taiwan MG Taiwan GO Taiwan GO Japan Total = 16,282.6

289.1

R15

130.9 1,559.8 1,650.9

3,210.7

2,677.8 1,312.3

130.9

806.8 392.4 959.7 2,519.6 356.3 5,034.7

3,990.1

3,627.0

Road route (unimodal): R1, R2, R3, R4, R9; intermodal route: R15. 2 RS = Rio Grande do Sul, MS = Mato Grosso do Sul, SP = São Paulo, SC = Santa Catarina, MG = Minas Gerais, PR = Paraná, MT = Mato Grosso, GO = Goiás. 3 Values in thousands of tons.

Brazil has a matrix of unbalanced transportation and logistical bottlenecks, hence, the costs of adapting the country’s infrastructure to the norms and standards of the CPB are higher in comparison to key competitors, the United States and Argentina, which have better logistics. The soybean and maize produced in Argentina travel shorter distances between production areas and ports of export, and the US prioritizes the waterway mode to distribute agricultural production. The scenario full segregation for maize clearly showed a negative impact on shipping, this is because approximately 70% of the intermodal route (more efficient in terms of cost, large volumes and long distances) become uncompetitive and are no longer used. In the case of soybeans, the situation is even worse since 77% of intermodal routes are no longer competitive. The implementation of full segregation systems greatly impact the Brazilian logistics making Brazil’s transportation matrix even more unbalanced. A full segregation system favors road freight transportation. However, the impact of the CPB in Brazil depends not only on the level of demand for segregation, but also on the compliance with the Protocol by the main importers. Importers should demand the same requirements from non-signatory countries, such as Argentina and the United States. If countries like Argentina and the United States do not have to follow the norms and standards set out by the CPB, Brazil may become even less competitive.

5. Final remarks The advances of the Brazilian agribusiness owe to a combination of factors, including more integrated supply chains, intensive capitalization in the various segments of the supply chains, and governmental support to agriculture. On the other hand, the logistics sector has been lagging behind, lacking adequate transportation infrastructure or storage facilities. The logistics of transport and storage, up until now marked by the movement of standardized products in large volumes, must adapt quickly to cope with the growing demand for differentiated and segregated products.

International Food and Agribusiness Management Review

De Oliveira and Alvim

Volume 20, Issue 1, 2017

Our findings suggest that trade flows required performing tests along the chain and resulted in the decreased competitiveness of the Brazilian maize and soybeans exports. The effect was greater in border states such as Mato Grosso. The requirement of segregation can interfere with the production decisions of farmers, without the production constituting a biosafety hazard. In a certain way, the CPB also imposes an increased opportunity cost when adopting a new technology. The monetary loss for soybean and maize was US$ 2.29 billion. This amount represents 12% of the foreign currency generated by the Brazilian exports of soybean and maize in 2011. The implementation of Identity Preservation Systems leads to an increase in fixed costs, with no direct connection to the fulfillment of the objectives of the Protocol, and may block the access of farmers to technology. It also negatively affects the competition among companies in the market of hybrid seeds, by delaying the release of cultivars resistant to insects and by limiting the offers available to farmers under the false argument that small farmers prefer local and non-hybrid varieties. The implementation of the CPB introduces a conflict between importers and exporters of agricultural commodities. On the one hand, importing countries attempt to establish an extremely demanding system on behalf of biosecurity. On the other hand, the large exporters of GMOs are concerned with the costs of implementing the Protocol and with new restrictions on international trade. Bilateral agreements and/or prediction of mechanisms to reduce tariffs imposed by importing countries may represent attempts to reduce the negative impacts of the CPB. Brazil is the second largest cultivator of biotech agriculture in the world and offers a complete and rigorous regulatory system. In the context of biotechnology activities, the Brazilian legal system is designed to protect consumers and the environment. Today, Brazil faces the challenge of reducing its deficit in transport and storage capacity. The country aims to increase its operational efficiency as well as take advantage of economies of scale and scope. The imposition of Identity Preservation Systems on a large scale would not only divert the necessary resources from agribusiness, but also creates uncertainty in the type of investment needed. Overall, it is critical that discussions on GMO regulation envision investments in infrastructure, so that advantages of agricultural biotechnology become available to more consumers in more countries.

References Ahumada, O. and J.R. Villalobos. 2009. Application of planning models in the agri-food supply chain: a review. European Journal of Operational Research 196: 1-20. Alvim, A.M. and P.D. Waquil. 2004. O problema de complementaridade mista: um modelo de alocação espacial aplicado ao setor agrícola. In: Métodos quantitativos em economia, edited by M.L. Santos and W.C. Vieira. UFV, Viçosa, Brazil, pp. 161-190. Bishop, P.M., C.F. Nicholson, J.E. Pratt and A.M. Novakovic. 2001. Tariff-rate quotas: difficult to model or plain simple? Paper presented at the annual conference of the New Zealand Agricultural and Resource Economics Society, Wellington, New Zealand. Available at: http://tinyurl.com/gnlebhl. Borges, I.C., J.M.F.J. Silveira and A.L.R. Oliveira. 2009. Constraints and incentives to agricultural biotecnology in brazil. Economia – Revista da ANPEC 10: 741-763. Brasil. 2006. Decreto nº 5.705, de 16 de fevereiro de 2006. Available at: http://tinyurl.com/jgzm3h8. Brooke, A., D. Kendrick and A. Meeraus. 1995. GAMS: a user´s guide. release 2.25. The Scientific Press, Redwood, OR, USA. Buanain, A.M. 2014. Alguns condicionantes do novo padrão de acumulação da agricultura Brasileira. In: O mundo rural no Brasil do século 21: a formação de um novo padrão agrário e agrícola, edited by A.M. Buainain, E. Alves, J.M. Silveira and Z. Navarro. Embrapa, Brasília, DF, Brazil, pp. 211-240. Companhia Nacional de Abastecimento (CONAB). 2014a. Armazenagem. Available at: http://tinyurl.com/ h9vswkz. Companhia Nacional de Abastecimento. 2014b. Levantamentos de safra. Available at: http://tinyurl.com/ zom22j4. International Food and Agribusiness Management Review

De Oliveira and Alvim

Volume 20, Issue 1, 2017

Dunn, T. 1994. Rapid rural appraisal: a description of the methodology and its application in teaching and research at Charles Sturt University. Rural Society Journal 4: 30-36. Food and Agriculture Organization (FAO). 2014. FAOSTAT. Availabe at: http://tinyurl.com/kzr77pz. Food and Agricultural Policy Research Institute (FAPRI). 2011. Elasticities database. Available at: http:// tinyurl.com/gra5wq7. Fuller, S., T.-H. Yu, L. Fellin, A. Lalor and R. Krajewski. 2001. Effect of improving South American transportation system on U.S. and South American corn and soybean economies (TAMRC International Market Research Report No. IM-2-01). Availabe at: http://tinyurl.com/j5yusnh. Fuller, S., T.-H. Yu, L. Fellin, A. Lalor and R. Krajewski. 2003. Transportation developments in South America and their effect on international agricultural competitiveness. Journal of the Transportation Research Board 1820: 62-68. Gruère, G.P. and M.W. Rosegrant. 2008. Assessing the implementation effects of the biosafety protocol’s proposed stringent information requirements for genetically modified commodities in countries of the Asia Pacific economic cooperation. Applied Economic Perspectives and Policy 30: 214-232. Huang, J., D. Zhang, J. Yang, S. Rozelle and N. Kalaitzandonakes. 2008. Will the biosafety protocol hinder or protect the developing world: learning from China’s experience. Food Policy 33: 1-12. International Service for the Acquisition of Agri-biotech Applications (ISAAA). 2015. GM approval database. Available at: http://tinyurl.com/hhzocle. James, C. 2013. Global status of commercialized biotech/GM crops: 2013. ISAAA Brief No. 46. ISAAA, Ithaca, NY, USA. Kalaitzandonakes, N. 2004. The potential impacts of the biosafety protocol on agricultural commodity trade. International Food and Agricultural Trade Policy Council. Washington, WA, USA. Available at: http://tinyurl.com/z5sx4s8. Mackenzie, R., F. Burhenne-Guilmin, A.G.M. La Viña and J.D. Werksman. 2003. An explanatory guide to the cartagena protocol on biosafety. Available at: http://tinyurl.com/j9pvttd. Oliveira, A.L.R., J.M.F.J. Silveira and A.M. Alvim. 2012. Cartagena protocol, biosafety and grain segregation: the effects on the soybean logistics in Brazil. E3 Journal of Agricultural Research and Development 2: 17-30. Rutherford, T.F. 1995. Extension of GAMS for complementarity problems arising in applied economic analysis. Journal of Economic Dynamics and Control 19: 1299-1324. Safras and Mercado. 2011. Banco de dados – Milho. Available at: http://www.safras.com.br. Samuelson, P.A. 1952. Spatial price equilibrium and linear programming. American Economic Review 42: 283-303.. Schlecht, S.M., W.W. Wilson and B.L. Dahl. 2004. Logistical costs and strategies for wheat segregation. Agribusiness and Applied Economics Report No. 551. North Dakota State University. Fargo, ND, USA. Sistema de Informações de Fretes (Sifreca). 2011. Fretes rodoviários e ferroviários – milho e soja. 2011. Available at: http://tinyurl.com/gtp2x8n. Simões, D.C. 2008. Regras, normas e padrões no comércio internacional: o protocolo de cartagena sobre biossegurança e seus efeitos potenciais para o Brasil. Master’s thesis, Universidade de São Paulo. Available at: http://tinyurl.com/jv2vcee. Takayama, T. and G. Judge. 1971. Spatial and temporal price and allocation models. North-Holland Publishing Company, Amsterdam, the Netherlands. Thompson, R.L. 1981. A survey of recent U.S. developments in international agricultural trade models. USDA/ERS. Available at: http://tinyurl.com/jmvkgxf. Thore, S.A. 1991. Economic logistics: the optimization of spatial and sectoral resource, production, and distribution systems. Quorum Books, New York, NY, USA. United States Department of Agriculture (USDA). 2011. Production, supply and distribution online. Available at: http://tinyurl.com/zn4ykur. United States Department of Agriculture (USDA). 2014. Production, supply and distribution online. Available at: http://tinyurl.com/zn4ykur. Waquil, P.D. and T.L. Cox. 1995. Spatial equilibrium with intermediate products: implementation and validation in the MERCOSUL. University of Wisconsin-Madison AAE Staff Paper No. 388. Available at: http://tinyurl.com/jxjp9tm. International Food and Agribusiness Management Review

OPEN ACCESS International Food and Agribusiness Management Review Volume 20 Issue 1, 2017; DOI: 10.22434/IFAMR2014.0099 Received: 11 August 2014 / Accepted: 7 November 2016

Predicting grower choices in a regulated environment RESEARCH ARTICLE Juan S. Castillo-Valeroa, Mercedes Sánchez-Garcíab, and Mari Carmen García-Cortijo aProfessor,

and cReseacher, Instituto de Desarrollo Regional, Universidad de CastillaLa Mancha, Campus Universitario s/n, Albacete 02071, Spain

bProfessor,

ETS Ingenieros Agrónomo, Universidad Pública de Navarra, Campus Arrosadía s/n, Pamplona, Albacete 31006, Spain

Abstract The analysis of farmers’ decision making process in the framework of agricultural policy is particularly complex as they take action within a structure of interacting opportunities, preferences, benefits and social factors which ultimately account for their behavior. This paper will study viticulturists’ behaviour vis-à-vis this scenario. Their decisions in the face of different possible alternatives are analysed using a multinomial logit model and a sample of 74,502 plots in Castilla-La Mancha (Spain). The conclusion is that viticulturists from this region are more prone to maintaining the status quo in their plots due to the current public support security issues, uncertain scenarios and their natural risk aversion. Keywords: agricultural policy, CMO, producer decisions, vineyard JEL code: Q12, Q18 Corresponding author: garcia.cortijo@gmail.com

Castillo-Valero et al.

Volume 20, Issue 1, 2017

1. Introduction The European Union regulation in the wine growing sector has experienced a number of changes and adaptations throughout the period of integration of the Common Market. The latest general change in general regulation for the wine sector dates from 2008; the Council Regulation (EC) No. 479/2008 (EC, 2008) repealed direct market intervention (distillations and grape must aid programs), which promoted temporary adjustment (grubbing up), as well as changes in vineyard planting rights regulation as from 2015. The specific case of this liberalization process was later reconsidered in the June 2013 political agreement on the reform of the Common Agricultural Policy (CAP) for the 2014-2020 period, which made planting restrictions possible through new decisions by administrative authorizations, though it put an end to their exchange rate in the transaction market, which completed the market freedom model with the end of distillations and the aforementioned structural adjustment. All this requires rigorous analyses in a context that allows for the discerning of the possible consequences this may cause. However, studies presenting an assessment of farm producers’ decisions in this context are still scarce. Added to that, farmers’ decision making is a complex process. Studies, such as Gasson (1973), Smith and Capstick (1976), Perkin and Rehman (1994), Sumpsi et al. (1997), Berbel and Rodríguez (1998), Costa and Rehman (1999), Willock et al. (1999), Solano et al. (2001), Bergevoet et al. (2004), Eastwood et al. (2012), Kanellopoulos et al. (2012), Leach et al. (2012) and Lybbert et al. (2012), all share the conclusion that when it comes to making decisions, farm producers take into account not only profit expectation but also the correct timing to make said decisions as well as a series of further considerations related to their economic, social, cultural and environmental context. Thus, the factors affecting farm producers’ decision making on managing options and the risks inherent to that process have generated an extensive literature in recent years (Moschini and Hennessy, 2001). Engler and Toledo (2010), Jones (2006), Moran et al. (2007) and Toledo et al. (2011) have pointed out the impact of farmers’ socio-demographic characteristics, such as educational level, age and gender (McRoberts et al., 2011; Nainggolan et al., 2013) on their decision making. An additional component is the degree of risk aversion of farm producers themselves, which is underlined by authors, such as Engler and Toledo (2010) and Girdžiūté (2012). Along with these decision makers’ characteristics, further structural factors are relevant, namely land ownership and membership in producer organizations (Engler and Toledo, 2010; Nainggolan et al., 2013). Economic determinants as well as those affecting business profits when it comes to making decisions are also important, as McRoberts et al. (2011), Moran et al. (2007), Nainggolan et al. (2013), Sattler and Nagel (2010) and Teschner et al. (2013) point out, as is receiving different types of subsidies (Nainggolan et al., 2013). That said, following Riesgo and Gómez-Limón (2006), farm producers will make decisions in such a way as to satisfy to the extent possible all their objectives taking into account all the relevant factors. Therefore, the main objective of this study is to analyze wine producers’ decision making process from a global perspective, taking into account various factors (structural, market, geographical, social) and the European regulatory policy framework (measured by aid received). In terms of this latter aspect, as stated by Garrido (2006) and Riesgo and Gómez-Limón (2006), farm producers are the actors who ultimately receive the corresponding programs and as such the success or failure of these programs depends on farmers. Subjects’ actions are taken within an opportunity structure interacting with their preference schemes, thus accounting for their behavior. In other words, whether or not farm producers decide to adhere to programs will be the result of the combination of their preference scheme (formed by their values and attitudes in relation to changes in agriculture and agricultural policy) and the structure within which they take action (Garrido, 2006). The wine sector is not immune to this process. In recent years, as a result of the entry into force of CAP regulations in the European Community through the 2008 Common Market Organisation (CMO), this sector has witnessed a normative adaptation which greatly affects farm producers’ decisions in relation to their holdings (among many other dimensions), especially those regulations related to structural aspects such as the management of potential production (planting management regime, permanent abandonment of wine-

International Food and Agribusiness Management Review

Castillo-Valero et al.

Volume 20, Issue 1, 2017

growing areas device or vineyard restructuring and reconversion aid scheme). Added to this, there are further structural, spatial, market and social determinants which also affect farmers’ decision making process. This research will present a logit multinomial model which, following Cabrer et al. (2001), will allow an analysis of economic agents’ behavior by capturing the level of probability of certain factors affecting their decision making process. This study specifically formulates a Logit Multinomial Model whose data comes from the Castilla-La Mancha region (Spain) and the material amounts to a sample of 74,502 plots with information provided by the Junta de Castilla-La Mancha 2012 Vineyard Register.

2. Theoretical framework Literature review Agricultural policy as public sector action – and public policy in general – have for decades been a decisive factor in the decision making process of farm producers and agents involved in the agri-food sector. Following Garrido (2006), the different actors in this sector, i.e. administrations, farming organizations and farmers themselves have been adopting positions on the matter and have transferred the regulation measures and instruments to the strategies they adopt. That said, it is true that farmers are the final beneficiaries and as such they decide to accede to any given specific program freely and on an individual basis. It is equally true that the decision making process is affected by multiple factors, some of them related to instrumental rationales (e.g. the appeal of direct aid) and others based on value-oriented rationales (e.g. the reduction of the negative impact of their technology model of choice). Profit maximization, the economic, social, cultural and environmental context and the timing of decision making should also be added to the list (Riesgo and Gomez-Limon, 2006). Research on farm producers’ decision making process in different sectors regulated by European agrarian policy has been a constant across the EU. In the dairy sector, highly affected by the quota system due to the delayed implementation of the quotas, Giannakas and Fulton (2000) show that farm producers take an opportunistic course of action in relation to agricultural policy measures. Their paper introduces misrepresentation and cheating into the policy analysis of output quotas and subsidies. Analytical results show that when cheating occurs output quotas are a less efficient means of income redistribution than is traditionally believed. Furthermore, cheating increases the transfer efficiency of output subsidies. The result is that an all-or-nothing choice between quotas and subsidies will generally favor the use of subsidies. A combination of quotas and subsidies, however, usually remains the most efficient means of income redistribution through market intervention. Helming and Peerlings (2002) also study the dairy sector and conclude that the abolition of the milk quota system in the Netherlands would result in dairy farmers increasing the number of milking cows. Jongeneel and Tonini (2009) conclude that farmers’ response capacities in terms of milk production is related to its price. The results in Kempen et al. (2011) show that if quotas were abolished, milk production in the EU would increase by more than 4%. Another study by Laepple and Hennessy (2012) notes that milk production depends of the real prices of milk. In terms of the sugar sector, also highly determined by the laying out of production quotas, the study by Nolte et al. (2012) concludes that farmers would increase production if the world market price went up. An analysis carried out by Rabobank (2013) shows that the abolition of sugar quotas in the EU in 2017 is expected to cause an increase in sugar production in the EU, which would also increase competition amongst suppliers. As to water, the basic input whose regulation is determining for farmers as they tailor the use of it according to said regulation, Jiménez et al. (2001) carried out an analysis of the impact that an increasing price of water would have in two irrigation zones. The comparison of the two communities studied show that irrigation aversion does not seem to be constant in farmers but rather has a clear relationship with property structure. Arriaza et al. (2002) conclude that irrigators’ behavior derives from the maximization of a utility function whose sole attribute is profit. Dinar and Saleth (2005) and Gómez (2009) conclude that the public provision International Food and Agribusiness Management Review

Castillo-Valero et al.

Volume 20, Issue 1, 2017

of water at subsidized prices has caused an increase in water consumption. Giannoccaro and Berbel (2011) point out that a CAP reform would have little impact on farmers’ decisions related to water usage due to the decoupling of aids to production. Finally, it is essential to note that economic, social and environmental dimensions do not have the same effects on farmers’ decision making (e.g. CAP agri-environmental measures). Garrido (2006) concludes that when it comes to farmers’ preference schemes in relation to agri-environmental policy, the overarching principle is an instrumental rationale; the programs are valued as opportunities to boost income. The economic dimension generally dominates farmers’ preferences whereas other dimensions, such as the social and environmental ones, have little or no impact. Therefore, although the economic dimension is indeed an important factor in decision making, the complexity of this process for farmers within the framework of public regulation and the different factors affecting them is nonetheless revealed as, following Freije and Rodríguez (1993), making a decision involves a reflection process which needs to take into account the pros and cons of the action alternatives and tries to opt for the more efficient one according to the objectives pursued. Now, as stated by Garrido (2006), taking a unidimensional approach which focuses solely on the economic dimension as the driving force in farmers’ decisions is insufficient in terms of accounting for the complex and heterogeneous agriculture reality seen as a space of production and sociocultural reproduction. If we bear in mind the increasingly multi-functional task attributed to farmers, it is our belief that a multi-dimensional approach (considering multiple factors) to their preference schemes is the most appropriate in the current context. In the case of the wine sector there are a few minor previous actions but so far exclusively related to the impact of the potential liberalization of planting rights (AEWR-UMR MOISA (2012), the European Parliament (2012), COPA-COGECA (2012) and the Report commissioned to the High Level Panel (2013). Furthermore, the paper by Deconinck and Swinnen (2013) provides a theoretical analysis of the economic effects and the social implications of planting rights. A model is proposed which takes into account land and production, trade restrictions and regional and national reserves. The model shows that liberalization creates winners and losers. Among the winners we find consumers, who benefit from larger wine supplies at lower rates. Owners of land other than vineyards also win due to the increase in land prices. A third group of winners are the new entrants in the sector, who will have the opportunity to plant vineyards. The losers are the owners of the original vineyards since the total value of their vineyards decreases and, furthermore, they face lower prices. Therefore, we believe it essential to carry out this research focusing on a multidimensional vision of farm producers’ reactions to public regulation measures within the framework of the wine CMO. Theoretical model As is the case with other agri-food sectors, public policy is a key element to understanding wine producers’ behavior when it comes to defining their management system. Throughout the 20th century the Spanish wine sector was subject to administrative regulations banning, promoting or regulating wine production, marketing and consumption. Then, when Spain signed the Accession Treaty to the European Community in 1986, the sector regulation and the CAP regulation became common to the twelve countries already in the European integration process by that year. Almost 30 years after that, the different actors in the European Community have been taking different positions, which are revealed in their discourses and strategies and in how they put them into practice through the corresponding policies (Garrido, 2006). It is thus essential to analyze and find out how the decision making process unfolds and study producers’ behavior in terms of probability to define, as this research does, the most likely decision that farmers will make. The decisions that farmers can adopt are strongly affected by European Union public regulation and are based on the premises in the 2008 CMO (EC, 2008), which sets out the guidelines for farmer behaviour. The deterioration of the balance between supply and demand in the wine sector, structural surpluses and lack of competitiveness in the sector justified the 2008 CMO. This regulation is based on four major objectives International Food and Agribusiness Management Review

Castillo-Valero et al.

Volume 20, Issue 1, 2017

(statement of intentions). The first one pursues improvement in competitiveness, promoting quality through the production of higher value added wines and the use of oenological practices. The second pillar is the control of the productive potential to achieve balance between supply and demand cutting out distillation as a way out for surplus production, and setting out a specific date to remove the restriction on planting rights, which is one of our main concerns in this study. The third generic instrument is market intervention through promotion in third country markets. A fourth block aims at strengthening the social fabric in rural areas and guaranteeing environmental protection. In June 2013, the establishment of administrative authorizations for new vineyard plantations was announced in the framework of the political agreement on the PAC reform for the 2014-2020 period to substitute for the current rights as of 2016 (EC, 2013a). This announcement again stresses the importance of assessing potential decisions on the part of producers, which is the main objective of our study. In an attempt to achieve all these goals in compliance with the regulations, a series of proposals were issued for farmers to choose from for their vineyards: (1) abandonment; (2) planting (provided they had planting rights or had acquired rights before the deadline for acquiring them and subsequent liberalisation); (3) grubbing up; and (4) restructuring. Articles 91 to 94 refer to planting and Chapter 3, to grubbing up (EC, 2008). Both options aim to achieve a balance between supply and demand of quality wine. The restructuring in Article 11 intended to increase producers’ competitiveness (EC, 2008). If local conditions in farmers’ plots are not conducive to viable production, they can also choose to abandon them. Article 68, which gives them an opportunity to cut costs and permanently withdraw these areas from wine production (EC, 2008). Faced with these options, the farmer can also choose to continue as usual (a 5th alternative). These five alternatives will make up a discrete dependent variable of the model described below. Thus, we can state that the main hypotheses (MH) analysed in this paper hold that the decisions made by winegrowers are a function of: MH1: structural characteristics of the vineyard (structural factors) MH2: price behaviour in the Spanish and international market (market factors) MH3: geographical location of the vineyards (geographical factors) MH4: land tenure and holding systems (social factors) MH5: payments received through the CAP measures (regulation) Structural factors (MH1) are directly related to land characteristics such as size, destination of production (table or quality wine), year of situation, management innovation and registration in the regulatory council. The expected effect of these variables will vary: (a) in terms of size, the smaller plots are more likely to be abandoned (Montagut and Gogliotti, 2008); and (b) destination of production. The process of planting and restructuring is more likely to take place in designation of origin or Protected Geographical Indication (PGI) areas (Basque Government, 2012); (c) innovation. Innovating plots are more dynamic and less prone to be abandoned (Giannoccaro and Berbel, 2011); and (d) Farmers’ union membership would favour planting and restructuring (Giannocarro and Berbel, 2011). The market variables (MH2) deemed adequate for the analysis are the reference price of wine and export price. Price is the basic factor here given that farmers’ income fosters vineyard development (HLP, 2013). Therefore, higher prices favor planting and restructuring and discourages abandonment and grubbing up. The geographical variables (MH3) analyzed are location in a Protected Designation of Origin (PDO) covered area and the population of the municipality where the plot is located. Geographical variables also need to be taken into account since unrestricted planting would cause the relocation of current vineyards International Food and Agribusiness Management Review

Castillo-Valero et al.

Volume 20, Issue 1, 2017

to more productive areas (HLP, 2013), and because of the risk of vineyard expansion beyond the delimited geographic PDO area (Basque Government, 2012). Therefore, the expected effect is an increase in plantations in delimited geographic PDO areas. As to social variables (MH4), another key element in the current vineyard scenario, following Engler and Toledo (2010) and Naiggolan et al. (2013), we include type of holding, type of owner and land tenure system. There is a concern that the number of family businesses will decrease and a more intensive, industrial wine growing will emerge in the context of a production model based on family holdings which sustains economic activity and employment in rural areas (COPA-COGECA, 2012; HLP, 2013). Based on this, family holdings will most likely tend towards abandonment and grubbing up whereas the opposite will be the case with industrial units. Finally, public regulation (MH5) should not be neglected. There are different opinions in this regard (Atance et al., 2001; Koráb, 2012) mainly because it raises producers’ expectations so they anticipate and accommodate their decisions and actions. Here we consider direct aid from the CAP as from the 2003 reform as regulation indicator. This variable could be expected to be either positive as in Cortignani and Severini (2011) and Giannoccaro and Berbel (2011) or hardly influential as noted by Nowicki et al. (2007; 2009). The theoretical model proposed in this research is summarized in Figure 1, which shows the framework within which farmers can act, the decisions they can adopt and the factors which can potentially affect them. The determining variables, as previously pointed out, have been divided into five large groups: Structural, Market, Geographical, Social and Public Regulation. These groups were inserted in the model estimate and will allow us to assess the degree of impact and significance of each variable and their groups in the Variables of grouped influence

Analysis of producer decisions

1. New planting 2. Restructuring/conversion 3. Grubbing up/one-time payment 4. Abandonment 5. Status Quo

A. Structural (MH1)

• Size of the plot • Destination of the production • Age • Innovation (variety + management) • Member of a Regulatory Council

B. Market (MH2)

• Reference grape prices • Export price

C. Geographical • Location/DO • Population (MH3)

Multinomial logit model

D. Social (MH4)

• Type of operator • Type of proprietor • Tenure regime

E. Public relations (MH5)

• CMO (distillations, restructuring and conversion, one-time payment, promotion)

Figure 1. Determinants in decision making by producers. MH1, MH2, MH3, MH4, MH5 represent the incidence variables in the producers decision that sustain the hypotheses. DO = designation of origin, CMO = Common Market Organisation. International Food and Agribusiness Management Review

Castillo-Valero et al.

Volume 20, Issue 1, 2017

face of the different options that emerge for farm producers in the new liberalising scenario: new planting, restructuring/converting, grubbing up, abandonment and status quo. The characterisation of producers, as expected, is particularly heterogeneous, as is the structure of their holdings. Therefore, it is hard to conceive that their behaviour would be similar to that of a standard farmer. Different behaviours amongst farmers are more likely to be expected and any change in their environment will tend to have a different impact on them (Pascual, 2007). Therefore, this study uses a micro-econometric model, specifically a multinomial logit model (MNLM). Its aim is to explain actual producer behaviour and the heterogeneity of their individual behaviours in a scenario regulated by the CAP and the CMO for wine.

3. Material and methods Sample and variables This paper uses data from wine-growers’ declared preferences in the Castilla-La Mancha Vineyard Register for each one of their plots (art. 8, 8/2003 Law of March 20th, on Vineyards and Wine from Castilla-La Mancha; GoE, 2003). The Vineyard Register is a tool for supporting the administrative management which considers the data relative to each holding. Both the ‘Estatuto de la Viña, del Vino y de los Alcoholes’ (Statute on Vines, Wines and Alcohols) and its Regulation (GoE, 1970) already considered in their articles 133 ff. in the constitution of a ‘Catastro Vitícola y Vinícola’ (Vineyard and Wine Register). At the EC level, the Council Regulation No 2392/86 of 24 July 1986 (EC, 1986) established a community Vineyard Register and the Commission Regulation No. 649/1987 (EC, 1987) laid down detailed rules for the establishment of said register. The region studied here is Castilla-La Mancha (Spain). However, following Recasens (2003) and Barco (2003), each wine region is susceptible to being analyzed through the same scheme; the results will be diverse but basically generalizable given the homogeneous behavior of producers as rational economic agents. But as Sartori and Robledo (2012) point out, models estimated only from declared preference data can lead to unrealistic predictions. Therefore, the sample was completed with revealed preferences (Brownstone et al., 2000; Page et al., 2000; Train and Wilson, 2008). The units of analysis are the plots rather than the producers given that the producers make different decisions for each of their plots as indicated by Jiménez et al. (2001). The database includes 74,502 plots out of the 617,071 plots in the Castilla-La Mancha Vineyard Register for 2012 (https://www.jccm.es). They were selected by stratified random sampling, the study plot population was divided into the groups corresponding to the alternatives of the endogenous variable, that is, abandonment (646), planting (7,296), grubbing up (23,518), restructuring (8,007) and status quo (35,035). A quota was assigned to each of these groups through proportional allocation according to the size of the population. Simple random sampling was carried out in each stratum so that all the plots would have the same probability of being selected, therefore preventing information bias. This reduction does not imply any significance problem for the results, firstly because it represents, as a whole, a sample error of 0.3% for finite mixtures. Individually, each of the categories of the dependent variable is also representative of the total population according to the sample errors that were obtained: 3.6% for abandoned plots, 1.1% for planted ones, 0.6% for grubbed up plots, 1% for restructured plots and 0.5% for those which remained unchanged.1 Second, it exceeds the minimum size of fifty observations marked by the asymptotic properties of maximumlikelihood estimators in a MNLM, (McFadden, 1974). It also surpasses the number of observations required by category and group of exogenous variables, established at a minimum of ten observations per exogenous variable in the endogenous variable category having the least representation (Schwab, 2012; Starkweather and

1 Abandoned (n=646 and N=5,452); planted (n=7,296 and N=56,497); grubbed up (n=23,518 and N=222,259); restructured (n=8,007 and N=60,402);

stayed the same (n=35,035 and N=272,234), in which n = sample size and N = population size.

International Food and Agribusiness Management Review

Castillo-Valero et al.

Volume 20, Issue 1, 2017

Kay, 2011). In our case, since there are 14 exogenous variables (13+ the independent term), there should be at least 140 plots. Yet our minimum number of plots in the least representative category (abandonment) is 646. Therefore, a cross section is formed that is suitable to be treated by the MNLM. The specified variables are shown on the following table (Table 1). The dependent variable consists of each of the alternatives which a vineyard farmer can opt for. These options are abandonment, planting, grubbing up, restructuring or staying the same, which are the regulatory instruments for vineyards in the EU established by the CMO for Wine (CMO, 2002). As shown on Table 1, it is a discrete variable which takes value 0 for plot abandonment; 1 for planting; 2 for grubbing up; 3 for restructuring; and 4 for status quo. The independent variables are diverse and have been classified as previously indicated into five groups: 1. Structural variables. This section consists of size, destination of production, year of action, innovation and inscription in the regulatory council. 2. Market variables. Here the price of grapes and the price of exports are included. 3. Geographical variables. These variables are classified into two groups: the areas of designation of origin and the population. 4. Social variables. These include type of owner, type of holding, and land tenure system. 5. Public Regulation reflected in CAP subsidies. Table 2 includes the descriptive statistics of the continuous exogenous variables and the frequencies of the discrete variables. The (mean) statistics on Table 2 (continuous variables: descriptive) show that within the action periods (AA) the year 1986 was the most dynamic, which was due partly to the expectations raised by the entry of Spain in the EEC (Arnalte, 2007). The mean reference price of wine (MRPW) has a mean value during the period under analysis of 2.54 euros/hectograde, a value slightly higher than the national mean. This difference widens during seasons of lower production. In the year 2012, there was a difference of 0.54 euros/hectograde more for Castilla-La Mancha (MAGRAMA, 2005). The mean unit price of wine exports (MUPWE) is 0.52 euros/ litre lower than the national mean for bulk wine sales. This difference has been increasing in recent years while Castilla-La Mancha has been adapting to the disappearance of distillations. In 2012 the price was 1.21 euros/litre in Spain, (OEMV, 2012). The mean population (MPOP) of the towns where the plots are located is close to 7,000 inhabitants; in Castilla-La Mancha 96% of all the towns have a population of under 10,000 inhabitants (INE, 2013), which in the sample represents 93.5%. Regarding CAP aid (MCAP), the provincial mean since 2000 is 58.5 million euros, 59% of the provincial mean for the rest of Spain according to data from the FEGA (FEGA, 2013). The analysis of Table 2 (discrete variables: frequencies) shows that farmersâ&#x20AC;&#x2122; majority decisions (CSITDE) after not changing the status quo (47%) were grubbing up (31%) and restructuring (11%). The size (NSUPER) of 66% of the plots is less than 10,000 m2. The production (CDESPR) from 72% of the vineyards was destined to QWpsr. Innovation techniques have been applied (INNOVA) in 11.65% of the plots, a number that coincides with the vineyards that underwent restructuring. Farmers registered in the regulatory council (CTPINS) reach 46% of the total. The plots located in a DO area come to 96.4%. In 95% of the cases, farmers who utilise the land (TIPEXP) are induvial. In 97% of the cases, the owners (TIPPRO) are also individuals. And regarding land tenure system (CREGTE), in 75% of the cases the land is their own property. Functional form of the model A MNLM was used to develop the research, as in papers by Geta et al. (2013), Ayuya et al. (2012) and Velandia et al. (2009), in which farmers choose the alternative J that gives them the greatest utility. In this research, the polytomous variable Y has five response categories that we named Y0, Y1, â&#x20AC;Ś,Y4 (Y0: abandonment; Y1: International Food and Agribusiness Management Review

Castillo-Valero et al.

Volume 20, Issue 1, 2017

Table 1. Table 1: Definition of the dependent and independent variables of the model logit. Variables

Typology

Description

Dependent CSITDE

discrete

Plot situation: 0 abandonment; 1 planting; 2 grubbing up; 3 restructuring; and 4 Status quo prepared by the authors using data from the 2012 JCCM Vineyard Register (Council of Castilla-La Mancha Communities)1.

Independent Structural NSUPERF discrete

CDESPR

AA INNOVA

CTPINS

Market MRPW

MUPWE

Geographical DO

MPOP Social TIPEXP TIPPRO CREGTE Regulatory MCAP

Plot surface area: 1 (≤10,000 m2); 2 (>10,000 and ≤100,000); 3 (>100,000). Prepared by the authors using data from the 2012 JCCM Vineyard Register (Council of CastillaLa Mancha Communities1). discrete Destination of production: 1 QWpsr; 2 Wine from the land; 3 Table wine. EC Regulation No. 479/2008 (EC, 2008). Prepared by the authors using data from the 2012 JCCM Vineyard Register (Council of Castilla-La Mancha Communities1). continuous Year in which the situation of the plot began. Prepared by the authors using data from the 2012 JCCM Vineyard Register (Council of Castilla-La Mancha Communities)1. discrete Innovation: 1=innovation; 0=no innovation. A plot is said to innovate when it has wiretrained vines and improved grape varieties. Prepared by the authors using data from the 2012 JCCM Vineyard Register (Council of Castilla-La Mancha Communities1). discrete Plot registered with the Regulatory Council: 1 if it is registered; 0 if it is not. Prepared by the authors using data from the 2012 JCCM Vineyard Register (Council of CastillaLa Mancha Communities1). continuous Reference price of grapes (€/hectograde) in the central region. The central region is limited to a series of towns. The remaining towns from the sample take the price of the closest town from the central region, the town least distant in kilometres. The variable is calculated as the average price of grapes from 2005 to 2012. Prepared by the authors using data from SEVI, 2005 to 2012 (Semana Vitivinícola, a winegrowers’ journal2). continuous Unit export price of wine (€/litre). Calculated as the quotient between the exported wine value and the volume. It is the mean provincial price from 2000 to 2012. Prepared from data from OEMV, 2012 (Spanish Wine Market Observatory3). discrete

Plot belonging to a Designation of Origin area: 1 if it belongs and (0) if not. Prepared by the authors using data from the 2012 JCCM Vineyard Register (Council of CastillaLa Mancha Communities1). continuous Average population of the town from 2000 to 2012. Data from INE (Spanish Statistical Office4) discrete discrete discrete

Type of operator: 1=individual; 2=legal entity. Type of owner: 1=individual; 2=legal entity. Tenure system: 1=ownership; 2=leased/sharecropping (prepared by the authors using data from the 2012 JCCM Vineyard Register) (Council of Castilla-La Mancha Communities1).

continuous Mean aid received in the period from 2000-2012 in euros. The mean was calculated for each province. Average calculated from data from FEGA (Spanish Agricultural Guarantee Fund5).

1 Available

at: https://www.jccm.es. at: http://www.sevi.net. 3 Available at: http://www.oemv.es/esp/-oemv.php. 4 Available at: http://www.ine.es. 5 Available at: https://www.fega.es. 2 Available

International Food and Agribusiness Management Review

Castillo-Valero et al.

Volume 20, Issue 1, 2017

Table 2. Descriptive statistics of the variables of the specified multinomial logit model.1 Continuous variables: descriptive

Obs. Mean Std. Dev. Min Max

MRPW

MUPWE

MPOP

MCAP

74,502 1,986 23 1,900 2,012

74,502 2.5444090 0.1847888 2.3709 3.39766

74,502 2.2297 0.2846211 0.9324641 2.429253

74,502 6,929.076 8,384.903 5.538 161,515.1

74,502 5.85e+07 3.99e+07 350,990 1.24e+08

Discrete variables: frequencies Code

CSITDE NSUPERF CDESPR INNOVA CTPINS

0 1 2 3 4 Σ

646 7,296 23,518 8,007 35,035 74,502

48,942 25,187 373

53,750 4,702 16,050

74,502

66,723 7,779

39,804 34,698

2,658 71,844

74,502

TIPEXP

TIPPRO CREGTE

70,674 3,828

72,300 2,202

55,585 18,917

74,502

AA= year; MRPW = reference price of grapes; MUPWE = unit export price of wine (€/litre); MPOP = mean population size; MCAP = received aid; CSITDE = plot situation; NSUPERF = plot surface area; CDESPR = destination of production; INNOVA = innovation; CTPINS = plot registered with the Regulatory Council; DO = Designation of Origin; TIPEXP = operator; TIPPRO = owner; CREGTE = tenure system.

planting; Y2: grubbing up; Y3: restructuring; and Y4: remaining unchanged). The aim was to explain the probability of each category depending on the group of observed co-variables X = {x1, x2, ..., xi},where i=13. That is to say, the aim was to adjust a model of the form pj (x) = P [Y = Yj | X = x] fj (x) ∀ j=0, ..., 4, for each vector x of observed values of the explanatory variables X. Therefore, the estimated formulations will provide a set of probabilities for the five alternatives (J+1) from which a farmer having X individual characteristics can choose. The covariables follow a multinomial distribution with probability parameters from each of the response categories, (Y | X = x) → M (1; p0 x), ..., pk (x)), where the sum of probabilities is one: k

∑ pj (x) = 1.

j=0

To construct the MNLM, (k–I) logit transformations were considered, defined as depending on a reference category, in this case Y4. Therefore the generalised logit transformations were defined as: pj (x) Lj (x) = ln ∀ j=0, ..., 3 p4 (x)

[

]

where Lj(x) is the logarithm of the response advantage Yj. Therefore, the model for each of the transformations is the following: 13

Lj(x) = ∑ βsj xs = x’ βj s=0

∀ j=0, ..., 3, for each vector of values observed from the explanatory variables x=(x0, x1, x2, ..., x13)′ where x0=1 and bj=(b0j, b1j, ..., b13j)′ are the parameter vector associated to the Yj category. The β coefficients are estimated by the maximum-likelihood method. After estimation, the model will be validated by means of the Likelihood Ratio Test and the pseudo coefficients of determination (McFadden, Cox&Snell, Nagelkerke, Count R2). Finally, the significance of the variables International Food and Agribusiness Management Review

Castillo-Valero et al.

Volume 20, Issue 1, 2017

was analysed jointly and individually. A group validation was obtained, as pointed out by Long (1997), with the Wald statistic and the LR test. The individual significance of each independent variable was analysed with the P-value associated with the z-distribution. The impact of each explanatory variable on the dependent variable will be interpreted through the marginal effects. Marginal effects have to be accounted for separately for each category of the dependent variable. The STATA 12 econometric software (StataCorp LP, College Station, TX, USA) was used to obtain statistical and econometric results.

4. Results This section shows the results obtained from the estimated MNLM, with a sample of 74,502 observations. The observations disaggregated by alternatives are: abandonment 646 plots; planting 7,296 plots; grubbing up 23,518 plots; restructuring 8,007 plots, and remaining unchanged 35,035 plots. First, the model was validated and then the significant variables in the model were ascertained. Regarding the validation of the model, the estimation of the multinomial logistic regression was obtained after nine iterations using the Newton-Raphson method. On the whole the model is significant with a probability associated with the Global Likelihood Ratio Test of zero (Prob>chi2=0.0000). This result is supported by the fit indicators: R2Mc =0.60 (excellent fit); R2Cox&Snell=0.773 (high fit, near the upper boundary; (ln LË&#x153; 0)2/N =0.91; R2Nagelkerke =0.845 (a value close to 1). Besides, the MNLM provides a 70% higher prediction level (Adj Count R2) than the highest frequency of the sample. Therefore, in 70% of the cases the prediction derived from the logistic regression model would be right (Table 3). Furthermore, the goodness of the model is verified by the likelihood ratio estimation (Table 3). The number of cases correctly predicted, 62,763 total plots, appears along the main diagonal of the matrix. It was 84.24% of the sample, a number that shows the goodness of the model. We continued with the combined significance of the model through the Likelihood-ratio and Wald tests, tests for independent variables. As observed on Table 3, both tests showed similar results (Long, 1997) and rejected the null hypothesis that the coefficients of the exogenous variables are simultaneously equal to zero since, with a confidence interval of 100, all variables were significant. Finally, the MNLM was validated with the Hausman test of independence of irrelevant alternatives (IIA) (Table 3). All five alternatives have negative coefficients, which, according to Hausman and McFadden (1984), is common for this type of tests. They conclude that this shows the necessary evidence that the independence of irrelevant alternatives was not violated, so that the null hypothesis of IIA was likewise accepted. In conclusion, it verified that the model is well specified. Finally, we focused on the individual estimate and significance (Table 4). We analysed the results of the estimation using the categories of the dependent variable: abandonment, planting, grubbing up, restructuring and status quo, this last being used as the base category. Abandonment With regard to abandonment, the smallest plots (NSUPERF) are most prone to it due to their lower level of profitability. Plots intended for production of table wine (CDESPR) are the ones that have experienced most abandonment; these being the most common in the region. At present (AA) abandonment of a plot is the least likely option. Innovation and technical change (INNOVA) are not present in the abandoned plots. Normally abandonment is linked to plots of unirrigated land, a long useful life and with head-pruned vines. The biggest drop occurs in seasons of low grape prices (MRPW) and therefore a less profitable crop, a result that has a clear effect on the short term expectations of the producer. By contrast, the price of wine in the international market (MUPW) is less influential in the decision to abandon. Being in a DO area has no significant influence due to its limited ability to generate significant added value based on territorial International Food and Agribusiness Management Review

Castillo-Valero et al.

Volume 20, Issue 1, 2017

Table 3. Diagnoses – evaluation model. Log likelihood=-36,257.766 (iteration 9) Base outcome=4 Log-Lik intercept only: -91,429.826 Log-Lik Full model: -36,257.766 LR chi2(50)=110,344.12; Prob>chi2=0.0000 Pseudo R2=0.6034 ML (Cox-Snell) R2: 0.773; Cragg-Uhler (Nagelkerke) R2: 0.845 Count R2: 0.842; Adj Count R2: 0.703 Tests for independent variables1,2

Likelihood-ratio chi2

P>chi2

chi2

P>chi2

NSUPERF CDESPR AA INNOVA CTPINS MRWP MUPWE DO MPOP TIPEXP TIPPRO CREGTE MCAP

177.926 2,910.332 72,316.893 19,597.566 2,910.732 135.753 40.395 576.684 88.550 68.442 98.915 670.112 149.390

4 4 5 5 4 4 4 4 4 4 4 4 2

0 0 0 0 0 0 0 0 0 0 0 0 0

176.672 1,719.674 14,768.996 6,937.924 2,237.418 135.356 37.196 348.053 88.763 64.897 97.936 621.35 136.869

4 4 4 4 4 4 4 4 4 4 4 4 4

0 0 0 0 0 0 0 0 0 0 0 0 0

Wald tests

Hausman tests of IIA assumption (N=74,502)3 Omitted

chi2

P>chi2

evidence

0 1 2 3 4

-97.216 -1.2e+03 -541.783 -119.002 -1,094.697

33 33 33 33 31

– – – – –

Likelihood ratio. Correctly predicted cases pred_ choice 0 1 2 3 4 Total

Total

6 1 3 2 634 646

1 559 4,918 1,076 742 7,296

0 102 22,591 77 748 23,518

0 301 1,803 5,680 223 8,007

35 90 694 289 33,927 35,035

42 1,053 30,009 7,124 36,274 74,502

1 NSUPERF = plot surface area; CDESPR = destination of production; AA= year; INNOVA = innovation; CTPINS = plot registered

with the Regulatory Council; MRPW = reference price of grapes; MUPWE = unit export price of wine (€/litre); DO = Designation of Origin; MPOP = mean population size; TIPEXP = operator; TIPPRO = owner; CREGTE = tenure system; MCAP = received aid. 2 H : all coefficients associated with given variable(s) are 0. 0 3 H : odds (outcome-j vs outcome-K) are independent of other alternatives; N = average population size. 0

International Food and Agribusiness Management Review

Castillo-Valero et al.

Volume 20, Issue 1, 2017

Table 4. Multinomial logistic regression results (number of observations=74,502).1,2,3

NSUPERF CDESPR AA INNOVA CTPINS MRPW MUPWE DO MPOP TIPEXP TIPPRO CREGTE MCAP Cons

Alternative 0 abandonment

Alternative 1 planting

Alternative 2 grubbing up

Alternative 3 restructuring

-0.3777215*** (0.1399474) 3.09756*** (0.1516839) -0.046084*** (0.0955732) -1.32855** (0.7236203) 0.0080845 (2.18e-09) -2.38149*** (0.4788038) 0.2016144* (0.1194986) 0.5675078*** (0.1220812) -0.0000128 (0.1516839) 0.0812888 (0.4562126) 0.0741402 (0.5225937) -1.77612*** (0.1853053) 1.01e-08*** (0.0000139) 85.74217*** (4.745932)

-0.0004644 (0.039878) 0.0498769** (0.024188) 0.1860118*** (0.0021109) 2.10372*** (0.08734) -0.0380567 (0.0420916) 0.3327845*** (0.0983089) -0.3066293** (0.1498377) 1.91491*** (0.2258258) 0.0000224*** (2.82e-06) -0.1754093* (0.1091514) 0.2104598* (0.1328757) 0.0365987 (0.0414341) 1.10e-08*** (1.05e-09) -374.58970*** (4.237612)

0.1752252*** (0.0383997) -0.5615843*** (0.0241528) 0.2881561*** (0.0024812) -2.46045*** (0.1572926) -0.0812698** (0.0403093) -0.3232281*** (0.1011016) -0.6495867*** (0.1359153) -1.04461*** (0.1366729) 0.0000172*** (2.83e-06) 0.4599622*** (0.1022304) -0.2321033* (0.1306718) -0.6604744*** (0.0415891) 5.69e-09*** (9.68e-10) -570.81290*** (4.956957)

0.5060231*** (0.0482814) 0.1807192*** (0.029624) 0.1832233*** (0.0026866) 5.14439*** (0.0887803) 2.28329*** (0.0584437) 0.5118017*** (0.115763) 0.1644401 (0.2250356) 2.61186*** (0.2492616) 9.21e-06*** (3.12e-06) -0.0116506 (0.1286273) -0.9780544*** (0.1626933) -0.084878* (0.0521772) 6.96e-09*** (1.49e-09) -373.23300*** (5.424165)

Standard errors are in parenthesis. denotes significance at the 10, 5, and 1% level, respectively. 3 NSUPERF = plot surface area; CDESPR = destination of production; AA= year; INNOVA = innovation; CTPINS = plot registered with the Regulatory Council; MRPW = reference price of grapes; MUPWE = unit export price of wine (â&#x201A;Ź/litre); DO = Designation of Origin; MPOP = mean population size; TIPEXP = operator; TIPPRO = owner; CREGTE = tenure system; MCAP = received aid. 2 *, **, ***

differentiation of quality. Plot owners in Castilla-La Mancha are very reluctant to leave (CREGTE). The option of definitively abandoning the cultivation of vines with or without obtaining financial aid or replanting rights is the least likely, since it would mean a capital loss (MCAP). Planting With regard to the planting category, the results of the model show that the size of the plot (NSUPERF) does not have excessive influence when it comes to carrying out planting, possibly because it is not so geared towards increasing competitiveness, nor it is associated with financial support or administrative restrictions. As for the destination of the production (CDESPR), the importance of table wine prevails due to its better performance in terms of yield and adaptation of the grape variety. Over the most recent years (AA), more dynamic planting, associated with a greater degree of innovation (INNOVA) was observed. Higher grape prices (MRPW) and the higher export price (MUPW) are significant in making the decision International Food and Agribusiness Management Review

Castillo-Valero et al.

Volume 20, Issue 1, 2017

to plant, contributing to an increased profitability of the crop. Planting takes place mostly in DO areas, due to the wide area they cover coupled with the potential added value that production there could acquire. The plots located in municipalities with the largest populations (MPOP) are more dynamic when it comes to planting tasks. Producers in these municipalities have greater competitive advantages over other areas since their increased economic activity results in increased availability of resources, infrastructure and services, such as access to wineries and input supplies. Planting is promoted by the owners of the plots (TIPPRO), and more by companies than by individual owners, because they usually have more entrepreneurial culture and economic resources. Grubbing up With regard to grubbing up, it is larger plots (NSUPERF) that are most prone to it because either renewing the vines or switching to another crop allows for greater profitability. The plots that produce table wine (CDESPR) are least likely to be grubbed up because most of regionâ&#x20AC;&#x2122;s wine is produced to this end and as a result of the expertise with which this is done it produces the best domestic results. The plots where the least innovation occurs (INNOVA) are the most grubbed up. Being registered with the Consejo Regulador (Regulatory Board) (CTPINS) does not induce the grubbing up of plots, since although the majority destination of production is table wine, wines that are finally classified as DO and bottled as such have higher added value. The price of grapes (MRPW) is crucial the decision to grub up, the result obtained being that the lower the price the more the grubbing up. The export price (MUPW) influences the decision to grub up since the international market is the main destination for the table wine production and this has expanded greatly in recent years. The municipalities with the biggest populations (MPOP) show greater dynamism in terms of grubbing up because producers have greater competitive advantages. The reticence towards grubbing up on the part of producers in Castilla-LaMancha (CREGTE) is contributed to by their receiving certain economic measures that involve income maintenance (MCAP). In this regard there are also certain intangible factors in play related to cultural and family values, combined with a lack of profitable crop alternatives. Regarding those who manage the land (TIPEXP), companies grub up more than individuals because they usually have more entrepreneurial culture and the economic resources to restructure the business and/or make more profitable use of it. Restructuring Restructuring is carried out on larger plots (NSUPERF), which allow producers to increase productivity by achieving economies of scale. Plots producing table wine (CDESPR) are the ones which have most undergone restructuring. This is because of the need to find an outlet for the large volumes of wine which previously ended up being distilled and which have now to be exported in bulk, without DO, PGI or having the grape variety identified. The plots in DO areas have a positive correlation with restructuring. This is because the majority of the plots in the region are located in DO areas, though not all wine produced there receives the DO label, as is evident from the fact that a large part of the regionâ&#x20AC;&#x2122;s production is table wine. Innovation and technological development (INNOVA) are present in the restructured plots. In part, this is due to the application of norms which regulate financial assistance for the restructuring of vineyards. Market conditions, particularly the price of grapes (MRPW) are critical to the decision on restructuring. Thus, a price increase directly encourages the decision to restructure since this system of cultivation enables a boost to yield along with lower production costs. Also regarding market conditions, the export price on the international market (MUPW) is less important than the grape prices when it comes to restructuring. Plots in municipalities with the largest populations (MPOP) show greater dynamism with regard to restructuring due to their greater competitive advantages. Wine producers in Castilla-La Mancha (CREGTE) are reluctant to restructure due to the costs involved and, especially, due to the limitations on irrigation. More owners (TIPPRO) take the decision to restructure than renters (TIPEXP). There are more owners than renters or other types of ownership of vineyards. CAP (MCAP) financial assistance is an important stimulus for restructuring. We complete these results with the information provided by the marginal effects (Table 5).

International Food and Agribusiness Management Review

Castillo-Valero et al.

Volume 20, Issue 1, 2017

Table 5. Multinomial logistic regression marginal effects (number of observations=74,502).1,2

NSUPERF CDESPR AA INNOVA CTPINS MRPW MUPWE DO MPOP TIPEXP TIPPRO CREGTE MCAP y 1 *, **, ***

Alternative 0 abandonment

Alternative 1 planting

Alternative 2 grubbing up

Alternative 3 restructuring

Alternative 4 status quo

-0.0000707** 0.000547*** -0.0000153*** -0.0001948*** -0.0000134 -0.0004249** 0.0000462*** 0.000072*** -2.88e-09 0.0000138 0.0000164 -0.0003065*** 1.47e-12 0.00017569

-0.0028089 0.0073592*** 0.0162302*** 0.0637108*** -0.013529*** 0.0330704*** -0.02745** 0.1003191*** 2.10e-06*** -0.0199295** 0.0256674** 0.0076283** 1.03e-09*** 0.11183567

0.0074532*** -0.0270093*** 0.0122395*** -0.0641998*** -0.0080629*** -0.0178348*** -0.0291254*** -0.093655*** 6.69e-07*** 0.0226292*** -0.0105458* -0.0311279*** 1.96e-10*** 0.04949867

0.0153123*** 0.0062552*** 0.0045277*** 0.6541154*** 0.0893657*** 0.0151008*** 0.0071769 0.0319352*** 1.77e-07* -0.0004595 -0.0305056*** -0.0016944 1.66e-10*** 0.03130968

-0.0198859*** 0.0128479*** -0.0329822*** -0.6534315*** -0.0677604*** -0.0299114* 0.0493524*** -0.0387609** -2.94e-06*** -0.002254 0.0153676 0.0255004** 1.46e-09*** 0.80668028

denotes significance at the 10, 5, and 1% level, respectively.

2 NSUPERF = plot surface area; CDESPR = destination of production; AA= year; INNOVA = innovation; CTPINS = plot registered

It should be noted that an increase of 1% in the size of the plot has a high impact on vineyard restructuring probability (1.5% increase) and on status quo (1.9% fall). The effect is not significant on any of the remaining alternatives. Moreover, large, more professionalized producers are revealed to be more prone to modernizing their holdings. Plots destined for table wine production are the ones that have been most often restructured and least often grubbed up. The importance of table wine still predominates in new plantings (according to the results of marginal effects, the likelihood of planting increasing is 1.6% with every passing year). The most recent years are the ones when more movement has taken place toward planting, grubbing up and restructuring. Abandonment is currently the least probable option. With every passing year, new planting is the option most likely to occur (1.6%) whereas the likelihood of maintaining the status quo in the plots decreases (-3.3%). Therefore, the consolidation of holdings and a stable regulation framework favor structural changes. Innovative plots were less prone to being abandoned or grubbed up. Rather, there was innovation in those which are newly planted or restructured. According to marginal effects, innovating farmers’ are most likely to decide for restructuring (65%). Being registered in the Regulatory Council stimulated restructuration (8.9% according to Table 5) and the opposite occurred for grubbing up and maintaining the status quo (-6.7%). With regard to market variables, the higher the price for grapes, the more motivation for farmers to plant and restructure and the less to abandon and grub up. The alternative most likely to occur after a 1% increase in price is planting, with a 3.3% increase. The status quo in the plots can also change by almost 3% after an increase in price. This reveals the importance of the economic variable as a determining factor in farmers’ decisions regarding their holdings. The price of wine on the international market was less important than the price of grapes in farmers’ decision making. Grubbing up and the status quo are worth highlighting; according to marginal effects a 1% increase in price results in a 2.9% decrease in grubbing up and a 4.9% increase in the status quo. This effect reveals a clear trend in farmers to take exporting as a profitable point of reference and as the pillar of the economic viability of holdings, underlined by the fact that the ones who focus on international markets when it comes to making decisions regarding their holdings are the most active and dynamic farmers. In terms of geographic variables, the majority of the sample plots are located within a DO area (96.4%; Table 3). Planting and restructuring are predominant as opposed to grubbing up. According to marginal effects, the probabilities of planting and restructuring increasing in PDO areas are

International Food and Agribusiness Management Review

Castillo-Valero et al.

Volume 20, Issue 1, 2017

10 and 3% respectively. This shows that a territorial reference label guarantees production and commercial invigoration and encourages farmers to take action. Plots located in towns with large populations are more dynamic in terms of planting, grubbing upand restructuring. On the other hand, in the least populated towns there was a higher probability of abandonment. With regard to social variables, farmers in Castilla-La Mancha are very reluctant to grub up and restructure, owners being the ones who make that decision. From the viewpoint of the agents who profit from the holdings, legal entities who are not land owners are the ones who promote grubbing up. As to regulatory variables, aid from the CAP is an important element in farmers’ decision making, which is revealed by its significance for all the alternatives (except for abandonment according to marginal effects; planting is the course of action of choice when aids increase, which reveals the high impact of subsidies on the decision to grub up and start new planting). The analysis of the influential variables in each of a producer’s alternatives led to the creation of the following plot categories: abandoned, planted, grubbed up and restructured plots. An abandoned plot corresponds to a small plot (<10,000 m2) with no innovations, whose grapes are destined for table wine, it located in small towns, grape price is low and has received subsidies. In a newly planted vineyard the size of the plot is irrelevant, there is innovation, the destination of production is table wine, the size of the town is irrelevant, grape prices are high and aid is important. Legal entities are more prone to planting. Grubbing up takes in large plots of where there is no innovation, the production is for table wine, the location is in larger towns, market prices are low (for grapes as well as exports) and aid is received. As for owners, neither individuals nor legal entities are inclined to grub up, particularly legal entities. Finally, a restructured plot is a large plot which innovates, grapes are destined for table wine, it is located in more highly populated towns and market prices for both grape and exports are high. Similarly to grubbing up plots, all types of owners are reluctant to choose this alternative. Summing up, the weight of each variable affects producers’ decisions, but the final decision is derived from the interaction of the whole of these variables, and not so much from the specific individual weight of any one of them. It is true that there are variables that turned out to be significant in all the alternatives. Destination of production (CDESPRO), year of action (AA) and innovation (INNOVA) were significant in structural variables; reference price (MRPW) in market variables; belonging to a DO area (DO) in geographical variables; and aid from the CAP in regulatory variables. Ultimately, farmers decisions are not easy to predict; hence the complexity of the analysis we have carried out. Then, we assessed the most probable alternative for a plot in Castilla-La Mancha using (1) all the plots from the sample, and (2) a standard plot whose characteristics are the average mean value of each of the exogenous variables of the sample. For all the plots from the sample (74,502), the most probable alternative was to remain unchanged with a mean probability of 0.47 (Table 6). For the standard plot, the option of remaining unchanged was also the most likely one with a probability of 0.8067. Therefore, the results in relation to the variables that affect winegrowers’ decisions were similar to the results obtained in previous research. Giannoccaro and Berbel (2011) studied how the CAP reform would affect farmers’ decisions on the use of water and they noted the importance of considering multiple factors simultaneously in their choices. Structural factors related to the size of the holding, interest in adopting innovations and belonging to agricultural groups, such as DO are specifically worth noting. Market and economic variables also affect decisions, in this case mainly through prices. Geographical and social variables are noteworthy depending on the location of producers. Finally, aid and subsidies received by the winegrowers were also important. Therefore, when it comes to designing agricultural policies, economic aspects have to be considered, but social and environmental ones are also important because of the effects they have on agricultural production activities, in this particular case on winegrowing.

International Food and Agribusiness Management Review

Castillo-Valero et al.

Volume 20, Issue 1, 2017

Table 6. Prediction of the probability of occurrence of each of the farmerâ&#x20AC;&#x2122;s decision alternatives: abandonment; planting; grubbing up; restructuring and status quo. Alternative1 0

Predicted probabilities for all plots in the sample obs 74,502 74,502 74,502 74,502 74,502 mean 0.0086709 0.0979303 0.3156694 0.1074736 0.4702558 std. dev. 0.0377504 0.1053838 0.3664378 0.2444321 0.442035 min 5.50e-11 5.38e-10 1.19e-13 3.14e-10 0.0002215 max 0.7516627 0.8278624 0.9948151 0.9837361 0.9999548 Predicted probabilities for a standard plot, while all other variables are held constant at their mean probability 0.0002 0.1118 0.0495 0.0318 0.8067 2 [0.0001; 0.0003] [0.1063; 0.1174] [0.0461; 0.0529] [0.0291; 0.0345] [0.7995; 0.8139] 95% CI 1 2

0 = abandonment; 1 = planting; 2 = grubbing up; 3 = restructuring; 4 = status quo. CI = confidence intervals by delta method.

5. Discussion Based on the analysis and results of the estimation, the conclusions contribute new elements that are complementary to most previous papers regarding potential structural imbalances. Regardless of the structural characteristics of the plots, the basic variables which affect producersâ&#x20AC;&#x2122; decisions are market variables such as the price of grapes and regulation variables such as CAP budgetary aid for grubbing up and restructuring and conversion (R&C). However, from the market variables viewpoint, the cost of exporting wine is not necessarily that important since producers do not take it as a direct reference in the decision strategy for their vineyards. In this study no relationship is observed between the price of exportation and the development of vineyards in the EU, including in Castilla-La Mancha. On the other hand, in countries outside of the EU supply is closely related to the market and tends to adjust to variations in price. The inexistence of a significant relationship could be due to the measures applied in the EU under the CMO for wine. From the structural viewpoint, there is a clear dichotomy between the actions of large and small holdings, as indicated above in the MAGRAMA report (2005). The former are much more inclined to modernise (R&C) (Gallego, 1996) but they also give priority to grubbing up, whereas small holdings have a greater probability of being abandoned. The same thing happens in the Autonomous Community of Navarre, where large holdings have higher survival rates than small ones (Aldanondo and Casanovas, 2009). This is what is generally the case in the agri-food sector in other geographic contexts as well (Montagut and Dogliotti, 2008). Furthermore, there is also a dominant structural variable with a clear tendency towards the concentration of vineyards in those areas with a denser amount of wine crops. It is in the areas featuring more vineyards where producers have decided to invest in modernising their vineyards and in new plantations. Studies of the Rheinland-Pfalz region in Germany (Bogonos et al., 2012a,b) and on the Priorat region in Catalonia, Spain (Bove, 2012) produce arguments along these same lines. The objective is to take advantage of the wine-growing potential in these areas. Innovation plays an important role in these new plantations and especially in restructured and converted holding. Producers feel that new plantations should be accompanied by changes and innovations to improve their competitive position, as established in the Council Regulation No 479/2008, Article 11, item 3 (EC, 2008). The same thing has happened in the community of Castilla-Leon

International Food and Agribusiness Management Review

Castillo-Valero et al.

Volume 20, Issue 1, 2017

in the Toro Designation of Origin (Sanchez, 2003), in the Aragon region (Government of Aragón, 2010) and in the wine producing sector of Priorat (Bove, 2012). On the other hand, the R&C process is probabilistically closer to taking place in areas having a DO and/or PGI than in areas where there is no territorial denomination. The same process happened in Aragon (Government of Aragón, 2010). Therefore, this geographical variable, as well as market positioning, affect producers’ decisions. However, new plantations have also been significant in terms of the potential tendency to choose these areas for their location. These new plantations, according to the report by the Basque Government (2012), would interfere with the quality objectives pursued by DO areas, since part of the objectives are obtained through a control of the vineyard surface area as well as of the conditions of production.

6. Conclusions This paper studied winegrowers’ decision making processes in Castilla-La Mancha in relation to public regulation in matters of structural changes in vineyards. The results show that farmers are faced with a complicated, ever-changing scenario and that their decisions are not easily predictable. In this scenario, the typical farmer from Castilla-La Mancha is generally more likely to adopt the decision of not carrying out any action in their vineyards. Grubbing up comes in second, restructuring and planting are found farther behind while the likelihood of abandonment is practically zero. Giannoccaro and Berbel (2011) also state that the majority of EU farmers declared their intention not to introduce any changes in their holdings. However, the model has shown that the provision of public aid and guaranteed prices stimulates investment on the part of farmers to expand their production capacity, exactly as Winter (2002) also suggests. As pointed out in the results analysis, the marginal effects studied reveal interesting conclusions in terms of winegrowers’ behavior; the economic variables focused on price (both grape price and average export price) are decisive for farmers to decide on modernizing their holdings. Along the structural dimension, larger holdings are more prone to opt for adding technological advances and changing production methods to reduce unit costs (restructuring and reconversion). Differentiation in production origin is an important element to prevent farmers from deciding to abandon and grub up vineyards, which validates the European DO and geographical indication labelling as a key element of wine policy in relation to other regions in the world. Finally, we have observed that the incentive of public aid favors farmers’ decision to move forward and decide on abandonment and grubbing up or new planting. Therefore, is would be safe to conclude that the effects of this potential measure, coeteris paribus the other factors and other EU actors, could affect farmers’ income through prices and this would increase the probability of abandonment and grubbing up. Taking into account that the probability of abandonment in Castilla-La Mancha is practically null and that of grubbing up is low, the effects of this potential measure are much more limited than what could be deduced a priori from a structural measure of this type, mainly due to the high probability of Castilla-La Mancha farmers maintaining the status quo in their plots (with a probability of 0.8067 according to the logit results). Therefore, agricultural policies have different effects and different degrees of impact because they are conditioned, as has been observed, by structural, market, geographic and regulatory variables. The results by Giannoccaro and Berbel (2011) are along the same lines. Therefore, this paper has aimed at improving research on farmer choices – a quite complex but at the same time highly appealing issue – both from the methodological and experimental point of view (Beckford and Barker, 2007; Bigliardi and Dormio, 2009; Fortuin and Omta, 2009). Furthermore, general patterns of behavior can be anticipated using the same methodology and an interesting line of future research is opened on the potential effects on producers’ decisions that the inclusion of the so-called authorisations for vine planting (EU Regulation No. 1308/2013 (EC, 2013b)) would cause in the wine growing sector to substitute for previous rights in the CAP reform for the 2015-2020 period. International Food and Agribusiness Management Review

Castillo-Valero et al.

Volume 20, Issue 1, 2017

References Aldanondo, A. and V. Casanovas. 2009. Análisis espacial del abandono de explotaciones agrarias en Navarra. Revista Española de Estudios Agrosociales y Pesqueros 222: 73-101. AREV-UMR MOISA. 2012. Etude sur les impacts socio-économiques et territoriaux de la libéralisation des droits de plantations viticoles. Available at: http://tinyurl.com/zsrumrf. Arnalte, E. 2007. Políticas agrarias y ajuste estructural en la agricultura española. Serie estudios, Ministerio de Agricultura. Pesca y Alimentación, Madrid, Spain. Available at: http://tinyurl.com/zbmlsad. Arriaza, M., J.A. Gómez-Limón and M. Upton. 2002. Local water markets for irrigation in southern Spain: a multicriteria approach. Australian Journal of Agricultural and Resource Economics 46: 21-43. Atance, I., I. Bardají and C. Tió. 2001. Política agrícola y competitividad. Efectos de sistemas alternativos de ayudas. Economía Agraria y Recursos Naturales 1: 111-124. Ayuya, O., W.S. Kenneth and G. Eric. 2012. Multinomial logit analysis of small-scale farmers’ choice of organic soil management practices in Bungoma County, Kenya. Current Research Journal of Social Sciences 4: 314-322. Barco, E. 2003. El mundo del vino. Available at: http://tinyurl.com/zw2qu4n. Basque Government. 2012. Effects of the liberation of vineyards in the autonomous community of the Basque Country. Ministry of the Environment, Territorial Planning, Agriculture and Fisheries, Spain. Beckford, C. and D. Barker. 2007. The role and value of local knowledge in Jamaican agriculture: adaptation and change in small -scale farming. Geographical Journal 173: 118-128. Berbel, J. and A. Rodríguez. 1998. An MCDM approach to production analysis: an application to irrigated farms in Southern Spain. European Journal of Operational Research 107: 108-118. Bergevoet, R.H.M., C.J.M. Ondersteijn, H.W. Saatkamp, C.M.J. Van Woerkum and R.B.M. Huirne. 2004. Entrepreneurial behaviour of Dutch dairy farmers under a milk quota system: goals, objectives and attitudes. Agricultural Systems 80: 1-21. Bigliardi, B. and A.I. Dormio. 2009. An empirical investigation of innovation determinants in food machinery enterprises. European Journal of Innovation Management 12: 223-242. Bogonos, M., B. Engler and S. Dabbert. 2012a. A Markov chains analysis for the growth of wine farms in Rheinland-Pfalz. Available at: http://tinyurl.com/j9ndkzl. Bogonos, M., B. Engler and S. Dabbert. 2012b. How the liberalization of planting rights will affect the wine sector in Rheinland-Pfalz, Germany: a partial equilibrium analysis. AAWE Working Paper 115. Bové, M.A. 2012. Innovation and internationalization in the wine sector in Priorat. Available at: http:// tinyurl.com/zz2e8xb. Brownstone, D., D. Bunch and K. Train. 2000. Joint mixed logit models of stated and revealed preferences for alternative fuel vehicles. Transportation Research Part B: Methodological 34: 315-338. Cabrer, B., A. Sancho and G. Serrano. 2001. Microeconometria y decisión. Ed. Pirámide, Madrid, Spain. Common Market Organisation (CMO). 2002. Evaluation of the common market organisation in the wine sector. Availabe at: http://tinyurl.com/zhqx3ep. COPA-COGECA. 2012. El papel de los derechos de plantación de cara al futuro del sector vitícola europeo. Available at: http://tinyurl.com/gozo76s. Cortignani, R. and S. Severini. 2011. An extended PMP model to analyze farmers´ adoption of deficit irrigation under environmental payments. Spanish Journal of Agricultural Research 9: 1035-1046. Costa, F.P. and T. Rehman. 1999. Exploring the link between farmers’ objectives and the phenomenon of pasture degradation in the beef production systems of Central Brazil. Agricultural Systems 61: 135-146. Deconinck, K. and J. Swinnen. 2013. The economics of planting rights in wine production. European Review of Agricultural Economics. Available at: http://tinyurl.com/gs5qkau. Dinar, A. and M. Saleth. 2005. Can water institutions be cured? A water institutions health index. Water Science and Technology: Water Supply 15: 17-40. Eastwood, C.R., D.F. Chapman and M.S. Paine. 2012. Networks of practice for co-construction of agricultural decision support systems: case studies of precision dairy farms in Australia. Agricultural Systems 108: 10-18.

International Food and Agribusiness Management Review

Castillo-Valero et al.

Volume 20, Issue 1, 2017

Engler, A. and R. Toledo. 2010. An analysis of factors affecting the adoption of economic and productive data recording methods of Chilean farmers. Ciencia e Investigación Agraria 37: 101-109. European Commission (EC).1986. Council regulation (EEC) No 2392/86 of 24 July 1986 establishing a community vineyard register. Official Journal of the European Communities L 208: 1-4. European Commission (EC).1987. Commission regulation (EEC) No 649/87 of 3 March 1987 laying down detailed rules for the establishment of a community vineyard register. Official Journal of the European Communities L 62: 10-17. European Commission (EC). 2008. Council regulation (EC) No 479/2008 of 29 April 2008 on the common organisation of the market in wine, amending Regulations (EC) No 1493/1999, (EC) No 1782/2003, (EC) No 1290/2005, (EC) No 3/2008 and repealing Regulations (EEC) No 2392/86 and (EC) No 1493/1999. Official Journal of the European Union L 148: 1-61. European Commission (EC). 2013a. The common agricultural policy after 2013. Available at: http://tinyurl. com/qjvdegt. European Commission (EC). 2013b. Regulation (EU) No 1308/2013 of the European parliament and of the council of 17 December 2013 establishing a common organisation of the markets in agricultural products and repealing Council Regulations (EEC) No 922/72, (EEC) No 234/79, (EC) No 1037/2001 and (EC) No 1234/2007. Official Journal of the European Union L 347: 671-854. European Parliament. 2012. The liberalisation of planting rights in the EU wine sector. Available at: http:// tinyurl.com/j4ea9pm. Fondo Español de Garantía Agraria (FEGA). 2013. Ayudas al sector vitícola. Available at: https://www.fega.es. Fortuin, F.T. and S.O. Omta. 2009. Innovation drivers and barriers in food processing. British Food Journal 111: 839-851. Freije, A. and S. Rodríguez. 1993. Control de gestión, optimización de las decisiones operativas. Ibérico Europea de Ediciones S.A., Madrid, Spain. Gallego, J.R. 1996. Instituciones, aprendizaje y liderazgo en la difusión de innovaciones. Una interpretación de la desigual implantación del riego por goteo en la citricultura valenciana. Revista Española de Economía Agraria 1: 199-226. Garrido, F. 2006. La cuestión ambiental en la agricultura: actores sociales y política agroambiental en España. Available at: http://tinyurl.com/hb7nhsk. Gasson, R. 1973. Goals and values of farmers. Journal of Agricultural Economics 24: 521-542. Geta, E., A. Bogale, B. Kassa and E. Elias. 2013. Determinants of farmers’ decision on soil fertility management options for maize production in Southern Ethiopia. American Journal of Experimental Agriculture 3: 226-239. Giannakas, K. and Fulton M. 2000. Efficient redistribution using quotas and subsidies in the presence of misrepresentation and cheating. American Journal of Agricultural Economics 82: 347-359. Giannoccaro, G. and J. Berbel. 2011. Influence of the common agricultural policy on the farmer´s intended decision on water use. Spanish Journal of Agricultural Research 9: 1021-1034. Girdžiūté, L. 2012. Risks in agriculture and opportunities of their integrated evaluation. Procedia-Social and Behavioral Sciences 62: 783-790. Gómez, M. 2009. La eficiencia en la asignación del agua: principios básicos y hechos estilizados en España. Economía y Medio Ambiente 847: 23-39. Government of Aragón. 2010. Analysis of the agri-food system in aragon: the wine sector. Available at: http://tinyurl.com/h3tcfrw. Government of España (GoE). 2003. Agencia estatal boletín oficial del estado. Ley 8/2003, de 20 de marzo, de la Viña y el Vino de Castilla-La Mancha. Available at: http://tinyurl.com/joapnr7. Go vernment of España (GoE). 1970. Agencia estatal boletín oficial del estado. Ley 25/1970, de 2 de diciembre, de Estatuto de la Viña, del Vino y de los Alcoholes. Available at: http://tinyurl.com/zs2fett. Hausman, J. and D. McFadden. 1984. Specification tests for the multinomial logit model. econometrica 52: 1219-1240.

International Food and Agribusiness Management Review

Castillo-Valero et al.

Volume 20, Issue 1, 2017

Helming, J. and J Peerlings. 2002. The impact of milk quota abolition on Dutch agriculture and economy, applying an agricultural sector model integrated into a mixed input-output model. In: X Congreso de la EAAE. Exploring Diversity in the European Agri-Food System. Zaragoza, Spain. Available at: http://tinyurl.com/h2djknt. High Level Panel. 2013. Report of the high level group on wine planting rights. Available at: http://tinyurl. com/gksxsqh. Instituto nacional de estadística (INE). 2013. Cifras de Población. Available at: http://tinyurl.com/hnzcmbk. Jiménez, J.F., J. Berbel and M. Torrico. 2001. Análisis de la toma de decisiones de los agricultores ante cambios en el precio del agua. Modelos de decisión multicriterio. Estudios Agrosociales y Pesqueros 190: 65-99. Jones, G.E. 2006. Modelling farmer decision-making: concepts, progress and challenges. Animal Science 82: 783-790. Jongeneel, R. and A. Tonini. 2009. The impact of quota rent and supply elasticity estimates for EU dairy policy evaluation: a comparative analysis. Agrarwirtschaft 58: 269-278. Kanellopoulos, A., P.B.M. Berentsen, M.K. van Ittersum and A.G.J.M. Oude Lansink. 2012. A method to select alternative agricultural activities for future-oriented land use studies. European Journal of Agronomy 40: 75-85. Kempen, M., P. Witzke, I. Pérez, T. Janson and P. Sckokai. 2011. Economic and environmental impacts of milk quota reform in Europe. Journal of Policy Modelling 33: 29-52. Koráb, P. 2012. European wine policy and perceptions of Moravian winemakers: theoretical background with an empirical study. MENDELU Working Papers in Business and Economics. Mendel University, Brno, Czech Republic. Available at: http://tinyurl.com/zy3w5ma. Laepple, D. and T. Hennessy. 2012. The capacity to expand milk production in Ireland following the removal of milk quotas. Irish Journal of Agricultural and Food Research 51: 1-11. Leach, M., J. Rosktröm, P. Raskin, I. Scoones, A.C. Stirling, A. Smith, J. Thompson, E. Millstone, A. Ely, E. Arond, C. Folke, and P. Olsson. 2012. Transforming innovation for sustainability. Ecology and Society 17: 11. Long, J. 1997. Regression models for categorical and limited dependent variables. Sage, London, UK. Lybbert, T.J., D.A. Sumner. 2012. Agricultural technologies for climate change in developing countries: policy options for innovation and technology diffusion. Food Policy 37: 114-123. McFadden, D. 1974. Conditional logit analysis of qualitative choice models. In: Frontiers in Econometrics, edited by P. Zarembka. Academic Press, New York, NY, USA, pp. 105-142. McRoberts, N, C., Hall, L.V., Madden and G. Hughes. 2011. Perceptions of disease risk: from social construction of subjective judgments to rational decision making. Phytopathology 101: 654-665. Ministerio de agricultura alimentación y medio ambiente (MAGRAMA). 2005. Programa de desarrollo rural para la mejora de las estructuras de producción en regiones fuera de objetivo N° 1 en España. Availabe at: http://tinyurl.com/h7zkn2g. Montagut, X. and F. Dogliotti. 2008. Alimentos globalizados: soberanía alimentaria y comercio justo. 2ª edición, Icaria Editorial, Barcelona, Spain. Moran, D., A. McVittie, D.J. Allcroft and D.A. Elston. 2007. Quantifying public preferences for agrienvironmental policy in Scotland: a comparison of methods. Ecological Economics 63: 42-53. Moschini, G. and D.A. Hennessy. 2001. Uncertainty, risk aversion, and risk management for agricultural producers. Handbook of Agricultural Economics 1: 88-131. Naiggolan, D., M. Termansen, M.S. Reed, E.D. Cebollero and K. Hubacek. 2013. Farmer typology, future scenarios and the implications for ecosystem service provision: a case study from south-eastern Spain. Regional Environmental Change 13: 601-614. Nolte, S., J. Buysse and G. Van Huylenbroeck. 2012. Modelling the effects of an abolition of the EU sugar quota on internal prices, production and imports. European Review of agricultural economics 39: 75-94. Nowicki, P., V. Goba, A. Knierim, H. van Meijl, M. Banse, B. Delbaere, J. Helming, P. Hunke, K. Jansson and T. Jansson. 2009. Scenar 2020-II – Update of analysis of prospects in the scenar 2020 Study. European Commission, Directorate-General Agriculture and Rural Development, Brussels, Belgium. International Food and Agribusiness Management Review

Castillo-Valero et al.

Volume 20, Issue 1, 2017

Nowicki, P., C. Weeger, H. van Meijl, M. Banse, J. Helming, I. Terluin, D. Verhoog, K. Overmars, H. Weshoek, A. Knierim, M. Reutter, B. Matzdorf, O. Margraf and R. Mnatsakanian. 2007. Scenar 2020 – Scenario study on agriculture and the rural world. European Commission, Directorate-General Agriculture and Rural Development, Brussels, Belgium. Observatorio Español del mercado del vino (OEMV). 2012. Principales importadores mundiales de vino. Available at: http://www.oemv.es/esp/-oemv.php. Page, M., G. Whelan and A. Daly. 2000. Modelling the factors which influence new car purchasing. European Transport Conference. PTRC, Cambridge, UK. Pascual, V. 2007. Modelos microeconométricos de elección en agricultura en condiciones de riesgo climático: aplicaciones al diseño de estrategias de adaptación al cambio climático. PhD thesis, Univ. Politécnica, Madrid, Spain. Perkin, P. and T. Rehman 1994. Farmers’ objectives and their interactions with business and life styles: evidences from Berkshire, England. In: Rural and farming systems analysis: European perspectives, edited by J.B. Dent and M.J. McGregor. CAB International, Wallingford, UK, pp. 193-212. Rabobank. 2013. Abolition of EU sugar quota. Available at: http://tinyurl.com/hvedhqh. Recasens, M. 2003. Economía vitivinícola en el siglo XXI. ACE Revista de Enología. Available at: http:// tinyurl.com/jmchetm. Riesgo, L. and J.A. Gómez-Limón. 2006. Multi-criteria policy scenario analysis for public regulation of irrigated agriculture. Agricultural Systems 91: 1-28. Sartori, J.J. and C.W. Robledo. 2012. Viajes al trabajo en la ciudad de Córdoba: estudio sobre la elección modal y la preferencia por la tenencia de vehículos. Revista Transporte y Territorio 7: 26-56. Sattler, C. and U.J. Nagel. 2010. Factors affecting farmers’ acceptance of conservation measures – a case study from North-Eastern Germany. Land Use Policy 27: 70-77. Schwab, J.A. 2012. Multinomial logistic regression: basic relationships and complete problems. Available at: http://tinyurl.com/jtd5wy4. Smith, D. and D.F. Capstick. 1976. Establishing priorities among multiple management goals. Southern Journal of Agricultural Economics Decisions 8: 37-43. Solano, C., H. León, E. Pérez and M. Herrero. 2001. Characterising objective profiles of Costa Rican dairy farmers. Agricultural Systems 67: 153-179. Starkweather, J. and A. Kay. 2011. Multinomial logistic regression. Available at: http://tinyurl.com/jk66aax. Sumpsi, J.M., F. Amador C. Romero. 1997. On farmers’ objectives: a multi-criteria approach. European Journal of Operational Research 96: 64-71. Teschner, N., Y. Garb and J. Paavola. 2013. The role of technology in policy dynamics: the case of desalination in Israel. Environmental Policy and Governance 23: 91-103. Toledo, R., A. Engler and V. Ahumada. 2011. Evaluation of risk factors in agriculture: an application of the analytical hierarchical process (AHP) methodology. Chilean Journal of Agricultural Research 71: 114-121. Train, K., W.W. Wilson. 2008. Estimation on stated-preference experiments constructed from revealedpreference choices. Transportation Research 42: 191-203. Velandia, M., R.M. Rejesus, T.O. Knight and J.S. Bruce. 2009. Factors affecting farmers’ utilization of agricultural risk management tools: the case of crop insurance, forward contracting, and spreading sales. Journal of Agricultural and Applied Economics 41:107-123. Willock, J., I.J. Deary, G. Edwards-Jones, M.J. McGregor, A. Sutherland, J.B. Dent, O. Morgan and R. Grieve. 1999. The role of attitudes and objectives in farmer decision making: business and environmentallyoriented behaviour in Scotland. Journal of Agricultural Economics 50: 286-303. Winter, M. 2002. Strong policy or weak policy? The environmental impact of the 1992 reforms to the CAP arable regime in Great Britain. Journal of Rural Studies 16: 47-59.

International Food and Agribusiness Management Review

OPEN ACCESS International Food and Agribusiness Management Review Volume 20 Issue 1, 2017; DOI: 10.22434/IFAMR2016.0017 Received: 21 January 2016 / Accepted: 19 October 2016

An application of activity-based costing in the chicken processing industry: a case of joint products RESEARCH ARTICLE Panravee Kabinlapata and Siriluck Sutthachai

aResearcher,

Faculty of Business Administration and Accountancy, Khon Kaen University, Khon Kaen, 40002 Thailand

bAssistant

professor, Accounting Department, Faculty of Business Administration and Accountancy, Khon Kaen University, Khon Kaen, 40002 Thailand

Abstract A great deal of research has been presented on the application of activity-based costing (ABC) in the manufacturing and service industries. In the field of agribusiness, which focuses uniquely on joint products, few studies exist that illustrate applications of ABC. This research has therefore applied ABC to a food company, concentrating on fresh and frozen chicken processing production. The basic process of ABC, based upon five steps, was applied and demonstrated the difficulties in applying cost data collection, identifying activity and cost drivers, as well as collecting driver data. The results also revealed significant differences in unit costs derived by ABC and the company’s existing cost system, particularly within frozen food products. This may suggest the possibility of distorted cost allocations within the company’s current costing system. Keywords: process manufacturing, activity-based costing, ABC, agribusiness, food company JEL code: M41 Corresponding author: sirsut@kku.ac.th

Kabinlapat and Sutthachai

Volume 20, Issue 1, 2017

1. Introduction Cooper and Kaplan (1988) introduced activity-based costing (ABC) which allocates overheads to end products or services based on the activities required for their production. Since its introduction, ABC has moved from concept to implementation (Tsai, 1996). Many researchers have discussed how to employ ABC in general and/or conceptual models (Carli and Canavali, 2013; Gunasekaran and Sarhadi, 1998; Roztocki et al., 2004; Schulze et al., 2011; Spedding and Sun, 1999), while others explored the use of ABC as case studies in manufacturing firms (Hughes, 2005; Jongprasithpron et al., 2006; Liu and Pan, 2007; Ong, 1995; Rezaie et al., 2008; Salehi et al., 2010; Tsai et al., 2012) and service businesses (Baykasoğlu and Kaplanoğlu, 2008; Themido et al., 2000; Vaughn et al., 2010). In the agricultural industry, Koutouzidou et al. (2015) reviewed studies on the application of ABC to several agricultural production systems; including fishing, winemaking, ornamental plant cultivating, sawmilling, and dairy farming, and concluded that there were limited applications within this industry, because its production methods have several unique characteristics that are not found in other industries. Of the research reviewed by Koutouzidou et al., (2015) González-Gómez and Morini (2006) applied ABC to winemaking processes that had no fixed operational procedures and in which the specific knowledge of the winemaker was crucial. In 2009, the same authors examined the application of ABC in multi-production of ornamental plants. However, current research has yet to catch up with the unique characteristics of joint processes, which are typically found in the food industry. The food industry produces a multitude of products from one or more primary materials through its manufacturing processes. That is, that joint products are simultaneously produced using a common or series of processes (Tsai, 1996). As such, this unique characteristic makes food companies an interesting case study for applying ABC. Tsai (1996) outlined a framework for applying ABC to joint products, but this concept has not been applied in practice. Therefore, this research applies ABC to a specific food company, and investigates whether ABC generates different product costs from the current cost system that is frequently used by many food manufacturing companies. This case study focuses on a food company in Thailand, as Thailand is an agricultural country with a high rate of food export, and a broad range of agricultural businesses. The Thai food industry has developed significantly and large manufacturing facilities have established many advanced technologies, in which production processes have become increasingly complicated. As a result, the indirect-cost proportion of the processes increases and must be accounted for in the unit cost calculation. Thailand also represents a developing country where ABC, as an advanced management accounting tool, has rarely been employed (Majid and Sulaiman, 2008).

2. The concept of ABC for joint products and an applied approach ABC is a costing system based on production activities. It assigns all overhead costs to activities with the assistance of resource drivers. Activity costs are then assigned to products according to activity drivers (Vazakidis et al., 2010). However, because of the unique characteristics of the joint production process in the food industry, there developed a need to modify standard ABC concepts. Tsai (1996) suggested that some joint products must pass through specific production process while others may be manufactured through an entire process. Thus, the second stage of cost allocation should be directed to each process, and then the costs of each process assigned to the joint products. According to the ABC framework and Tsai’s suggestion, this study follows the approach in Figure 1. The first step identifies resource costs and activities. Resource costs refer to the ingredients required in a production process, such as labor and materials, which cannot be directly assigned to particular products, and are often described as overhead (OH). Oliver (2000) defines activities as a description of the work necessary to produce a product in a company; which uses people, technology, materials, processes, and the International Food and Agribusiness Management Review

Kabinlapat and Sutthachai

Volume 20, Issue 1, 2017

1. Identifying resource costs and activities 2. Identifying resource and activity drivers 3. Assigning resource costs to each activity and process 4. Assigning activity costs to each product 5. Cost analysis

Figure 1. The ABC approach (adapted from Baykasoğlu and Kaplanoğlu, 2008; Nachtmann and Al-Rifai, 2004; Tsai, 1996). ABC = activity-based costing environment. This information can be obtained by reviewing the documentation of work flow, observing the actual production process, and interviewing employees (Baykasoğlu and Kaplanoğlu, 2008). The second step indicates resource and activity drivers, both of which are generally selected based on three criteria: (1) they are easy to identify, simple to use, and uncomplicated; (2) they have a direct relationship to indirect costs; and (3) they generate an advantage for understanding the behavior of costs, and influence the changing of cost (Gary and Sorinel, 2010). Several approaches were employed to identify the drivers; such as brainstorming, meetings, interviews, questionnaires, observations and/or combinations of approaches (Baykasoğlu and Kaplanoğlu, 2008; Themido et al., 2000; Tornberg et al., 2002). In the third step, resource costs are allocated according to the resource consumption coefficients of the resource drivers, and are calculated in proportions (in which the sum of the proportion in each driver=1). Several researchers (e.g. Baykasoğlu and Kaplanoğlu, 2008; Salehi et al., 2010) utilized the expense activity dependence (EAD) matrix to assign OH costs to specific activities. Due to joint process characteristics, the activities’ costs are then assigned to the processes. The fourth step involves allocating the process costs to the product. Similar to the method employed in the third step; the consumption coefficients are calculated and the activity product dependence matrix may be used when there are a wide range of processes and products. The final step compares the product cost based on ABC with that initially calculated by the company through other means, and analyzes similarities and/or differences.

3. Company background This case study1 is on a food corporation in Thailand that runs a complete supply chain of poultry and swine integration for both domestic and international markets. Its operations were classified into six types of business; regional and feed, poultry integration, swine integration, food integration, animal health, and other. For this study, the poultry integration business was chosen as the case because it is successful and creates the main revenue for the company. The poultry integration business encompasses chicken feed mills, breeder farms, hatcheries, broiler farms, contract farms, fresh and frozen chicken processing, cooked chicken products, and chicken meatballs and sausages. The products from fresh and frozen chicken processing represent our main focus, as their production processes are complex and justify the application of ABC. 1 The

company name is anonymous in order to prevent a potential competitive disadvantage.

International Food and Agribusiness Management Review

Kabinlapat and Sutthachai

Volume 20, Issue 1, 2017

The company’s fresh and frozen chicken products The products derived from fresh and frozen chicken processing are structured and outlined in Figure 2. There are six main parts: the carcass, fillet, bone-in leg, breast bone, wing, and whole chicken. Four of these chicken parts (bone-in leg, breast bone, fillet, and wing) can be further processed into several product variations. For example, the breast bone can be categorized into breast-bone meat and a skinless breast-bone meat. Each of these products can be further allotted to three sub-products: domestic, frozen and raw material. The other three primary parts can be separated into the products presented in Figure 2. Each product has three sub-products, except for the drumstick, which has two sub-products: domestic and raw material. Therefore, there are six primary parts, separated into 15 products, and further processed into 38 sub-products. These sub-products were grouped from 136 total products manufactured by the company. The production area and process The company has two production factories: the slaughter house and the cooked-food house. The former is in focus, for its processing of the fresh and frozen chicken products. The slaughter house area is divided into eight zones, illustrated in Figure 3: 1. ‘Live birds zone’ receives and weighs live chickens, and conducts a quality control procedure, an ante-mortem inspection. The chickens are then unloaded and transported on a conveyor belt that leads to the slaughter zone. 2. ‘Slaughter zone’ is for slaughtering chickens. This is gently performed by skilled employees. There are six steps in this zone: hanging and stunning, slaughtering, scalding, de-feathering, washing, and head and feet removal. The whole chicken is prepared for production, and the blood, head, feet, and feathers become by-products. 3. ‘Evisceration zone’ includes two machine-operated processes: (1) evisceration and post-mortem inspection – the chickens are eviscerated by the auto-eviscerating machine and veterinarians perform the post-mortem inspection; and (2) inside and outside washing – in which the whole chicken is placed in a washing machine. 4. ‘Chiller zone’ contains the chiller and sorting machines, which freezes and sorts chickens by weight. A whole chicken is passed through the chiller machine at 4 °C to preserve the freshness of the chicken. 5. ‘Primary cut-up zone’ cuts the chicken into five main parts: fillet, leg bone, breast bone, wing, and carcass. Four parts are sent to the secondary cut and special product separation zone. The carcass is sent to the storage process (Zone 8).

The chicken products (cost objects)

Live chicken

Whole chicken

Bone in leg

Thigh

Drumstick

Domestic

Breast bone

Meat of breast bone

Fillet

Skinless breast bone

3 join wing

Domestic

Frozen

Carcass

Wing stick

2 join wing

Meat of bone

Raw material

Wing

Wing tip

Middle wing

Raw material

Figure 2. Fresh and frozen chicken products (information obtained from the interviews and company documents). International Food and Agribusiness Management Review

Kabinlapat and Sutthachai

Live birds zone

Volume 20, Issue 1, 2017

Live chickens receiving and anti-mortem inspection

Scalding

Slaughter zone

Feather

De-feathering

Washing

Evisceration zone

Blood

Slaughtering

Hanging and stunning

Head and feet cutting

Head and feet

Eviscerating and post-mortem inspection -

Organs

Inside and outside washing Chiller zone

Whole chicken

Chilling and sizing Cut up

Primary cut up zone

Yakitori zone

Bone in leg Breast bone

Fillet

Carcass

Wing

Cut up the primary products for each product type

Post-by product

Inspection

By-product leg meat part

Cut and trim unwanted part

By-product leg bone part

Secondary cut up product and special product separation zone

By-product breast bone part

Select and classify product Pack and weight

By-product fillet part

Vacuum pack and metal detection Reducing the temperature

Freezing zone

Inspection the temperature Storage Storage zone Domestic

Raw material

Export

Figure 3. Production zones in the slaughter house (information obtained from the interviews and company documents).

International Food and Agribusiness Management Review

Kabinlapat and Sutthachai

Volume 20, Issue 1, 2017

6. ‘Secondary cut-up and special product separation zone’ include five processes: 6.1 Secondary cut: four parts from the primary cut-up zone are separated into sub-products by skilled laborers. 6.2 Inspection: laborers check the products by sampling whether the meat will satisfy customer quality requirements. If not, the meat is sent to the cut-and-trim process. 6.3 Cutting and trimming the unwanted parts: laborers cut and trim the unwanted parts, such as bruised skin, bones, blood, or scrap. These are by-products and sold in domestic markets. 6.4 Selection and classification: laborers sort the chicken into three groups: domestic, export, and raw material; based on features, such as weight and shape. Each group is transferred to its productline process. 6.5 Pack and weight: laborers pack and send the products to the appropriate departments. Domestic products are sent to an exposing department in the storage zone, raw materials are transferred to the chilling room, and exported products are delivered to the freezing zone. 7. ‘Freezing zone’ consists of three processes: 7.1 Vacuum packing and metal detection: meat is packed and sealed with a vacuum packing machine and inspected by the metal detection machine. 7.2 Freezing process: sealed packages are transferred to the freezing room and are immediately frozen at -18 °C to preserve their quality. 7.3 Quality control process: frozen products are randomly sampled to check whether their temperature is appropriate. 8. ‘Storage zone’ is for storing products in a different way. Domestic products are kept in a domestic warehouse, while raw material products are transported by trolley to a chilled room. Frozen products are transported to a warehouse that maintains a specific temperature. The company’s current cost allocation method The company currently uses the absorption costing method, which accumulates costs throughout the production process and allocates these costs to an individual product (Figure 4). It also employs the sales value at the split-off-point to allocate joint costs to joint products and by-products. However, the absorption costing method was not directly connected to the corporate accounting system. If the company wished to connect it, significant costs would be incurred. Therefore, production costs were calculated in an offline system. Cost data was gathered from the accounting system, and calculated as follows: 1. In the primary cut processes, the total joint costs consisted of all costs associated with four units: live bird, de-feathering, evisceration, and primary cut; which typically comprised of direct costs, production overhead, and service overhead. Some overhead costs were directly incurred in the unit, such as the unit manager’s salary; but other unidentified-unit costs, such as electrical expenses and the plant manager’s salary, were allocated to each unit based on selected drivers. 2. In each unit, when by-products were generated, the total joint costs were allocated to each by-product at the split-off point, based on its marketable value. The net total joint costs were then assigned to the six primary parts, based on their proportion of the total marketable value. The overall cost allocation process is presented in Figure 4. 3. After the primary cut processes, the allocated joint costs of the four main parts (bone-in leg, breast bone, fillet, and wing) were assigned to 136 sub-products. Each sub-product also absorbed its own additional production costs, such as the freezing expenses incurred by the frozen sub-products. The cost allocation for the secondary cut processes, seen in Figure 4, uses the fillet products as an example. In sum, the company’s cost allocation method allows by-products to absorb production costs according to their marketable value. Consequently, a by-product with a high marketable value would have higher joint costs, while one with a low marketable value would be assigned a low cost. This may not distort the production costs of each product if the by-products have a low market price and if they represent a small proportion of

International Food and Agribusiness Management Review

Kabinlapat and Sutthachai

Volume 20, Issue 1, 2017

Accumulated joint costs of 4 units (live bird, de-feathering, evisceration, cut-up) = direct costs + production and service overhead + allocated OH Allocated by marketable value

Costs of a whole chicken and by-products – blood, feather, heads and feet, organ, unwanted parts, such as bruising fillet.

Net total joint costs of 4 units

Allocated by the proportion of the total marketable value

A whole chicken

Bone in leg

Breast bone

Fillet

Wing

Carcass

Allocated by the marketable value

Main products

By-products

Allocated by the proportion of the total marketable value

Fillet tip offunsize

Fillet tip off

Fillet 5 kg

Fillet no. 2

Fillet frozen

+ Additional costs

...

Figure 4. The current cost allocation method that the company applied (information obtained from the interviews and company documents). OH = overhead. the company’s total production; otherwise, management could misunderstand the production costs, which could lead to unintended mistakes or decisions. Similarly, the current method allowed the joint costs assigned to each sub-product based on its proportion of total marketable value, which suggests that a sub-product with high marketable value could have high allocated joint costs, and vice versa. Therefore, it is possible that a product with uncomplicated production processes could have higher costs than those with complicated processes.

4. Results The results of the applied ABC approach are explained below. Identifying resource costs and activities We observed the production processes and conducted initial interviews with management from three departments: cost accounting, slaughter house, and cooked products. The company’s documents were reviewed, including the work flow of the slaughtering process, a list of products, and the chart of accounts. Eighty-eight accounts of resource costs were identified but due to their diversity, they were grouped into 26 overhead categories based on their similarities (Baykasoğlu and Kaplanoğlu, 2008). Work flows were analyzed and drafted for each type of product. The flow illustrated the process of production, previously presented in Figure 3. Each process consisted of various complicated activities; some of which could not be clearly identified. The research therefore identified activities according to their manufacturing processes.

International Food and Agribusiness Management Review

Kabinlapat and Sutthachai

Volume 20, Issue 1, 2017

A second interview was conducted with the company’s managers to confirm expense categories and activities. Some expenses were determined whether they should be allocated to production. After thoroughly consulting with the accounting department, some were classified as ‘non-allocated’ expenses because the transactions’ direct relation to production was unclear. All expense categories are shown in Supplementary Table S1. In deriving the activities, the production manager clarified the work flow of the production process in detail and suggested improvements, so that the activities were well-defined. As a result, 16 activities were identified, and are presented in Table 1. Identifying resource and activity drivers A discussion with management and consultation with the work flow document were conducted to indicate the resource and activity drivers, based on criteria outlined by Gary and Sorinel (2010). To allocate some expenses, more than one resource drivers could be employed. For example, the training expenses could be allocated to activities according to either the number of staff or the number of training courses. In this case, management suggested that most of the training courses were organized for all staff, and were not specific to particular departments. Therefore, the staff number was considered a more suitable resource driver of the training cost allocation than the number of training courses. The problem of alternative drivers for allocating costs also appeared when the activity drivers were identified. Therefore, at this step, drivers were identified strictly on the condition that they satisfied the criteria of Gary and Sorinel (2010) and management reviewed and approved them. The resource and activity drivers are outlined in the above tables. Assigning resource costs to each activity/process This step is to allocate OH to activities and then activity costs to processes. Because the research identified activities according to the manufacturing processes, cost allocation to processes was not needed. Supplementary Table S2 reports the consumption rates and costs of each activity in the EAD matrix, and demonstrates how much production cost each activity consumed. Before assigning costs to activities, the research collected data on the resource costs and drivers. Cost data was obtained from the accounting system, which primarily records transactions for financial reporting purposes; thus, some expenses included both direct and indirect costs, such as staff expenses, including the salaries of staff who work in a particular process (direct costs), as well as on various procedures (indirect costs). It is necessary to accurately identify the indirect costs when applying ABC. The company addresses this problem through the use of subaccounts to indicate indirect costs, and then manually collects those subaccounts for cost management. Our study relied upon this cost data. Resource-driver data were primarily collected from the slaughter house department; however, portions of the required data were not available, as the company had no data collection system. Therefore, other possible drivers were determined, and obtained within the second step: identifying resource and activity drivers. After obtaining the cost and driver data, consumption rates for each activity were calculated by dividing the amount of resources consumed by each activity with the total amount of resource consumption. For example, for vehicle expenses (No. 6 in Supplementary Table S2), the resource driver was the number of times a vehicle was used, which was 10,872 times. This total number contributed to four activities: marketing and selling (691 times), administration and support (10,062 times), waste water treatment (107 times), and customer service (12 times). As a result, the consumption rate for marketing and selling was 0.0636 (691/10,872). Administration and support consumed the most vehicle expense, as its consumption rate was 0.9255. Customer service and the waste water treatment activities consumed rates 0.0098 and 0.0011, respectively. Consumption rates for other activities were calculated in the same way. According to the consumption rates, the total amount of each expense was allocated to each activity. Continuing from the vehicle expense example above, the total vehicle expense of $460,476 was distributed to four activities, according to their consumption rates. Therefore, the marketing and selling activity accounted International Food and Agribusiness Management Review

Kabinlapat and Sutthachai

Volume 20, Issue 1, 2017

Table 1. Details of the activities and activity drivers. Activities/process

Explanation

Activity drivers

1. Marketing and selling 2. Ordering and planning

A set of activities performed to obtain orders from customers. A set of activities that uses the process of receiving an order from customers and scheduling a production plan according to the orders. 3. Slaughtering Activities of preparing a whole chicken to be ready to cut that includes receiving live chickens, stunning, slaughtering, scalding, de-feathering, washing, eviscerating and sizing the whole chickens. 4. Primary cut A group of activities that involve cutting. Chickens are put in the loop of head conveyor and passed through the breast bone, wing, leg bone, and fillet stations for cutting. The carcass is the last main part received. 5. Secondary cut A set of activities that include detail cutting and trimming the breast bone, wing, leg bone and fillet. 6. Checker weigh A set of activities that record skilled laborers’ performance in cutting and trimming particular products. Their performance depends on the weight of chickens that are recorded to a card or machine and then sent to the administration. 7. Packing and A group of activities that includes selecting the shaped weighting parts of chicken, their weight, and their quality, according to customers’ specifications, and the packing. 8. Vacuum packing and Two main activities: (1) vacuum packing, to pack metal detection selected parts of the chicken with a special technique that is performed by a machine to keep the meat from spoiling; and (2) metal detection, the machine detects unwanted pieces in the packages. 9. Chilling activity The activity is to maintain the products at a temperature of 0-4 °C. 10. Freezing The activity is to freeze the product by quickly lowering its temperature to the appropriate degree. 11. Storage An activity for storing the finished frozen products before transferring them to customers or sending them to be a ready-to-eat product. 12. Domestic products A group of activities that prepare products that are sold to domestic customers. 13. Delivery-out A group of activities for shipping frozen products to overseas customers by hiring a third party. 14. Administration and A set of activities that exist to support and facilitate staff support and laborers who work in manufacturing. 15. Waste water An activity for converting waste water to clean water treatment before released. 16. Customer services A series of activities designed to enhance the level of customer satisfaction.

no. of customers person no. of product types

type

product quantity kilogram

amount of time used in cutting

second

amount of time second used in cutting product quantity kilogram

product quantity kilogram

product quantity kilogram product quantity kilogram product quantity kilogram

product quantity kilogram product quantity kilogram no. of product type types product quantity kilogram no. of customers person

International Food and Agribusiness Management Review

Measurement unit

Kabinlapat and Sutthachai

Volume 20, Issue 1, 2017

for $29,286.27 (460,476 × 0.0636); administration and support activity received $426,170.54 (460,476 × 0.9255); waste water treatment and customer service activities received $506.52 (460,476 × 0.0011) and $4,512.66 (460,476 × 0.0098), respectively. Allocations of other expenses were calculated in the same way. Supplementary Table S2 provides the consumption rates and costs allocated to each activity. After allocating all costs, we summed the total costs for all activities. The slaughtering activity costs were the highest, at approximately 8.5 million dollars, and customer service activity costs were the lowest, at approximately $71,180. Assigning activity costs to each product At this stage, the total cost for each activity was allocated to the 38 product types. Supplementary Table S3 shows the activities associated to each product and several of their consumption rates. Wingtip-frozen and skinless breast-bone-frozen products consumed 14 activities in the production procedures whereas the whole chicken product consumed only five activities. To calculate the total OH cost for each product, its activity consumption was multiplied by the activity rate (Supplementary Table S2). Due to the many types of products, and to clearly present the differences between the current cost system and ABC, this paper exemplified the OH cost calculation of the fillet products, as illustrated in Supplementary Table S4. Product cost analysis Comparisons between the current cost system and ABC are presented in Table 2. In both the company’s current system and the proposed ABC concept, the frozen fillet products had the highest costs. However, in comparing these costs, the ABC cost was higher than that of the current system about $2.17 and $2.21 per kilogram. This difference suggests that the current cost system possibly has not allocated all associated OH costs to frozen fillets regarding their production activities. Table 2. A comparison of the product costs.1 No. Product types

1 ABC

domestic fillet fillet fillet No. 2 bruising fillet fillet 1 fillet 5 fillet (BF) fillet BFS frozen fillet fillet frozen fillet frozen 2 raw material fillet fillet unsize fillet fillet ML

ABC

The case study

OH per unit ($/kg)

Direct costs per unit ($/kg)

Total costs per unit ($/kg)

0.3299 0.3299 0.3299 0.3299 0.3299 0.3299 0.3299

1.0294 1.0294 1.0294 1.0294 1.0294 1.0294 1.0294

1.3594 1.3594 1.3594 1.3594 1.3594 1.3594 1.3594

1.9011 1.8826 1.7206 1.8143 1.8689 1.9174 1.7431

3.13 3.13

1.0294 1.0294

4.16 4.16

1.9580 1.9891

0.3257 0.3257 0.3257

1.0294 1.0294 1.0294

1.3551 1.3551 1.3551

1.8969 1.8880 1.8894

= activity-based costing; OH = overhead. International Food and Agribusiness Management Review

Diff.

-0.5418 -0.5232 -0.3612 -0.4549 -0.5095 -0.5581 -0.3838 2.21 2.17 -0.5417 -0.5329 -0.5343

Kabinlapat and Sutthachai

Volume 20, Issue 1, 2017

The product costs for the two other types of fillet products based on the ABC were lower than those from the company. This is consistent with the frozen fillet products, as the company may be allocating more OH costs to the other fillet product types rather than to the frozen products. Therefore, the two types of fillet products would have lower costs if their overhead costs were ABC based. Differences in production cost methods (company’s existing system and ABC) were found in other product types, sharing a similar explanation: frozen products have higher costs than other types of products, as a result of their assigned consumption activities.

5. Conclusions and recommendations The purpose of the study was to apply ABC in a food production company that had the uniqueness of joint processes. The Thai company operating a fresh and frozen chicken processing production was selected as the case study to achieve this objective. Its production generates more than one hundred products, assigned to 38 product groups. It was currently using an absorption cost system and the selling value split-off point method for its production costs. The study employed the five steps of ABC application, interviews, documentation, observation, and consultation for data collection and analysis; as well as an EAD matrix and an activity-product-dependence matrix, given the wide range of processes and product variations. The results from the applied ABC provided several conclusions. First, that identifying activities can be challenging, as the food industry’s manufacturing process is complicated and consists of many related activities. We addressed this difficulty by identifying activities according to production procedures, and consulting with various production managers. Additionally, there are problems with cost and activity data availability. Because most food manufacturing companies apply a process cost system, cost data is primarily accumulated through the flow of production. Some data had not been clearly indicated as either direct or indirect costs, such as staff salary. Any application of ABC would therefore require a separation of direct and indirect costs. Shigaev (2015) suggested that at least two groups of accounts should be created for recording related costs when applying ABC. And, for activity data, data collection should be modified if it is not available in the company’s system. Another consideration of importance in applying ABC to the food industry is the determination of resource and activity drivers. Some activities may be found to have more than one appropriate driver, which fairly allocate overhead costs; whereas other activities may be found to have no suitable cost drivers at all. Selection criteria must be precise and practical. Another concern in ABC application is that certain costs cannot be allocated to a particular activity (Spedding and Sun, 1999) and, thus the costs may be classified as nonallocated expenses. The difficulties in allocating resources to activities, and from activities to individual products were also presented by Zakić and Borović (2013). Although the research results demonstrated the benefits of ABC compared to the company’s existing cost system, this should be interpreted with caution, as the accuracy of product costs based on ABC depends upon numerous random factors, such as the selection of activities and cost drivers (Spedding and Sun, 1999). The ABC application presented in this paper sheds light on the possibility of distorted cost allocations within the industry’s current costing methods. Current methods are typically used in several food manufacturing companies for cost allocation; yet may not fully realize the production activities that each product passes through in the cost allocation procedure. The ABC approach could address this possible distortion and provide more accurate cost information for company management.

International Food and Agribusiness Management Review

Kabinlapat and Sutthachai

Volume 20, Issue 1, 2017

Acknowledgements This work was supported by the Higher Education Research Promotion and National Research University Project of Thailand, Office of the Higher Education Commission, through the Food and Functional Food Research Cluster of Khon Kaen University, Thailand. We also gratefully acknowledge the valuable information and comments provided by the company and all anonymous reviewers.

Supplementary material Supplementary material can be found online at https://doi.org/10.22434/IFAMR2016.0017. Table S1. The details for the 26 expense categories and resource drivers. Table S2. Assigning resource costs to the activities that are presented in the EAD matrix. Table S3. The activity product dependence matrix showing the amount of activity consumed by some products. Table S4. An example of the overhead cost calculation for the fillet products.

References Baykasoğlu, A. and V. Kaplanoğlu. 2008. Application of activity-based costing to a land transportation company: a case study. International Journal of Production Economics 116: 308-324. Carli, G. and M. Canavari. 2013. Introducing direct costing and activity based costing in a farm management system: a conceptual model. Procedia Technology 8: 397-405. Cooper, R. and R.S. Kaplan. 1988. Measure costs right: make the right decisions. Harvard Business Review 66: 96-103. Gary, C. and C. Sorinel. 2010. Cost drivers evolution and benefits. Theoretical and Applied Economics 8: 7-16. González-Gómez, J.I. and S. Morini. 2006. Activity-based costing of wine. Journal of Wine Research 17: 195-203. Gunasekaran, A. and M. Sarhadi. 1998. Implementation of activity-based costing in manufacturing. International Journal of Production Economics 56: 231-242. Hughes, A. 2005. ABC/ABM – Activity-based costing and activity-based management: a profitability model for SMEs manufacturing clothing and textiles in the UK. Journal of Fashion Marketing and Management 9: 8-19. Jongprasithpron, M., S. Limnararat, S. Jongprasithpron and N. Yodpijit. 2006. Activity-based costing: A case study of manufacturing motorcycle front and sprockets. (in Thai) The Journal of KMITNB 16: 1-6. Koutouzidou, G., A. Vazakidis, A. Theodoridis and C. Batzios. 2015. A review of ABC methodology for agricultural sector. Proceedings of the 7th International Conference on Information and Communication Technologies in Agriculture: Food and Environment (HAICTA 2015), Kavala, Greece, 17-20 September 2015. Available at: http://tinyurl.com/jpktdtw. Liu, L.Y.J. and F. Pan. 2007. The implementation of activity-based costing in China: an innovation action research approach. The British Accounting Review 39: 249-264. Majid, J.A. and Sulaiman, M. 2008. Implementation of activity based costing in Malaysia. Asian Review of Accounting 16: 39-55. Nachtmann, H. and M.H. Al-Rifai. 2004. An application of activity based costing in the air conditioner manufacturing industry. The Engineering Economist 49: 221-236. Oliver, L. 2000. The cost management toolbox: a manager’s guide to controlling costs and boosting profits. Amacom, New York, NY, USA. Ong, N.S. 1995. Manufacturing cost estimation for PCB assembly: an activity-based approach. International Journal of Production Economics 38: 159-172. Rezaie, K., B. Ostadi and S.A. Torabi. 2008. Activity-based costing in flexible manufacturing systems with a case study in a forging industry. International Journal of Production Research 46: 1047-1069. International Food and Agribusiness Management Review

Kabinlapat and Sutthachai

Volume 20, Issue 1, 2017

Roztocki, N., J.D. Porter, R.M. Thomas and K. L. Needy. 2004. A procedure for smooth implementation of activity-based costing in small companies. Engineering Management Journal 16: 19-27. Salehi, M., R. Hejazi and N.B. Manesh. 2010. Activity based costing model for cost calculation in gas companies: empirical evidence of Iran. International Review of Accounting, Banking and Finance 2: 31-43. Schulze, M., S. Seuring and C. Ewering. 2011. Applying activity-based costing in a supply chain environment. International Journal of Production Economics 135: 716-725. Shigaev, A. 2015. Accounting entries for activity-based costing system: the case of a distribution company. Procedia Economics and Finance 24: 625-633. Spedding, T.A. and G.Q. Sun. 1999. Application of discrete event simulation to the activity based costing of manufacturing systems. International Journal of Production Economics 58: 289-301. Themido, I., A. Arantes, C. Fernandes, and A.P. Guedes. 2000. Logistic costs case study – an ABC approach. Journal of Operational Research Society 51: 1148-1157. Tornberg, K., M. Jämsen and J. Paranko. 2002. Activity-based costing and process modeling for cost-conscious product design: a case study in a manufacturing company. International Journal of Production Economics 79: 75-82. Tsai, W.H. 1996. Activity-based costing model for joint products. International Conference on Computers and Industrial Engineering 31: 725-729. Tsai, W.H., Y. Shen, P. Lee, H. Chen, L. Kuo and C. Huang. 2012. Integrating information about the cost of carbon through activity-based costing. Journal of Cleaner Production 36: 102-111. Vaughn, P., C. Raab and K. B. Nelson. 2010. The application of activity-based costing to a support kitchen in a Las Vegas casino. International Journal of Contemporary Hospitality Management 22: 1033-1047. Vazakidis, A., I. Karagiannis and A. Tsialta. 2010. Activity-based costing in the public sector. Journal of Social Sciences 6: 376-382. Zakić, V. and N. Borović. 2013. Application of activity-based costing in agricultural enterprises. The seminar: agriculture and rural development – challenges of transition and integration processes. Department of Agricultural Economics. Belgrade-Zemun. Available at: http://tinyurl.com/z8fqhf4.

International Food and Agribusiness Management Review

OPEN ACCESS International Food and Agribusiness Management Review Volume 20 Issue 1, 2017; DOI: 10.22434/IFAMR2016.0003 Received: 6 January 2016 / Accepted: 10 September 2016

Food safety as a field in supply chain management studies: a systematic literature review RESEARCH ARTICLE Daniel P. Auler a, Rafael Teixeirab, and Vinícius Nardic aPhD

Candidate, Department of Management and Operations, Universidade do Vale do Rio dos Sinos (UNISINOS), Flores da Cunha 93/406, CEP 93.010-160, São Leopoldo, State of Rio Grande do Sul, Brazil bProfessor,

Department of Management and Operations, Universidade do Vale do Rio dos Sinos (UNISINOS), João Neves da Fontoura, 211/301, CEP 93.010-050, São Leopoldo, State of Rio Grande do Sul, Brazil cPhD

Candidate, Department of Management and Operations, Universidade do Vale do Rio dos Sinos (UNISINOS), Rua Livramento, 515, CEP 95.700-000, Bento Gonçalves, State of Rio Grande do Sul, Brazil

Abstract The increasing number of food contamination events has called the attention of both practitioners and scholars to food safety problems and their consequences. Many of these events are related to the supply chain because food production is now a global process of commoditized goods, made by large corporations that purchase inputs from producers in many countries. Given the linkage between supply chain and food safety issues, we investigated how studies in the supply chain management area have examined food safety issues, exploring some of their important characteristics. To do so, we conducted a systematic literature review of 46 papers, published in 23 journals, indexed in the Web of Science database. As a result, we pointed out some main characteristics of these papers, including journal attributes, authorship data, citation network, methodological characteristics, and theoretical approaches. Results serve as a reference to scholars and allow us to discuss some potential opportunities for future research in the field of food safety in the supply chain management area. Keywords: supply chain management, food safety, systematic literature review. JEL code: M11, L66, L23, Q56, M19 Corresponding author: auler.dani@gmail.com

Auler et al.

Volume 20, Issue 1, 2017

1. Introduction The food contamination events occurring in the last few years have called the attention of practitioners and scholars to better food safety practices. In 2008 and 2009, the United States and Canada found evidence that peanut butter paste was contaminated with salmonella, causing 9 deaths and affecting another 637 people, (Layton and Miroff, 2009). In 2008, pork goods produced in Ireland were contaminated by dioxin and affected international suppliers in several countries (EFSA, 2008). In China, milk powder was adulterated with melamine and was associated with 6 deaths and 294,000 contaminated people (Spencer, 2009). In 2011, bean sprouts produced in Germany were contaminated by E. coli, resulting in 37 deaths and another 3,000 people contaminated (Marucheck et al., 2011). These episodes and their consequences claim urgent and rigorous treatment, as they cause serious problems for public health and firms’ value and profits (ResendeFilho and Hurley, 2012; Roth et al., 2008). These contamination events are partially related to changes in modern supply chains, such as the globalization of food produce, consolidation of large companies, and commoditization of food products (Roth et al., 2008). These changes put pressure on producers to reduce cost (Tang and Babich, 2014), which, in turn, respond by reducing the investment in safety actions. Also, in global supply chains, the distribution centers play another essential role in helping the prevention of safety hazards, because they store food products and can detect food contamination before such food reaches the consumers (Chebolu-Subramanian and Gaukler, 2015; Lao et al., 2012). The number of tiers in the chain, for example, also influences the risk of food contamination, because the existence of many interacting elements increases the likelihood of chemical and physical contamination (Reiner and Trcka, 2004; Sloane and O’Reilly, 2013; Van der Gaag et al., 2004). Given the growing importance of supply chain elements in food contamination events and the increasing number of papers published about this topic, we seek to investigate how supply chain scholars have examined food safety issues by answering the following research questions: What are the main characteristics of studies about food safety in the area of supply chain management research? What are the aspects that describe the authors involved in these studies? Which methodological and theoretical features do these studies have? To answer these questions, we propose a systematic literature review to analyze a sample of papers published in respected management journals and to reveal their main characteristics. By answering these questions, we can contribute to the literature by summarizing how issues related to food safety have been explored by supply chain management scholars. In doing so, we hope to contribute by indicating some points that could be better designed in future researches, particularly aspects such as methodological characteristics, theoretical approaches, possible partnerships between scholars and institutions, country focus available, and journals for potential publication.

2. Background A supply chain can be viewed as a network or a group of firms interconnected by multiple buyer-supplier relationships in which products and information flow to help the process of developing, producing, and delivering goods and services to consumers. For instance, a firm could be the supplier or customer of other similar firms, selling or buying inputs for production activities. Because of these relationships, the supply chain management has emerged as one of the important concepts to help companies improve their performance through the use of better supply chain practices (Burgess et al., 2006). There are many studies related to different issues of supply chain management, all of which attempt to find ways to improve the performance of the entire supply chain (Marucheck et al., 2011; Pérez-Mesa and Galdeano-Gómez, 2015). Following this same line of reasoning, a food supply chain can be defined as a set of interdependent companies that manage the flow of goods, services, and information along the value-added chain of agricultural and food products seeking to achieve superior customer value at the lowest possible cost (Beske et al., 2014). A typical food supply chain is illustrated in Figure 1 (Roth et al., 2008). It is important to note that importers

International Food and Agribusiness Management Review

100

Auler et al.

Volume 20, Issue 1, 2017

Farm suppliers

Marketers/storage

Processors

Exports

Imports

Farmer

Wholesalers/distributors

Retailers

Consumers

Figure 1. Generic model of food supply chain (adapted from Roth et al., 2008). and exporters are linked to all phases of the supply chain system, showing that any tier of the chain can be connected to the global supply chain. Globalization of the food supply chain gives rise to some challenges to management, especially because this type of chain has become increasingly dynamic and industrialized (Roth et al., 2008). Some of these challenges are food traceability (Alfaro and RĂĄbade, 2009; Epelbaum and Martinez, 2014) and tracking (Fritz and Schiefer, 2009), supply chain dispersion (Rong and Grunow, 2010) and food distribution (Akkerman et al., 2010), quality assurance (Ting et al., 2014), risk assessment (Dani and Deep, 2010; Wang et al., 2012), and related trust and commitment among firms (Ding et al., 2014). One common characteristic of these studies is that they can be viewed from two perspectives: food security and food safety. Food security refers to the delivery of food â&#x20AC;&#x2DC;that is uncompromised by intentional contamination, damage, or diversion within the supply chainâ&#x20AC;&#x2122; (Marucheck et al., 2011: 708). Security problems can arise from other people or organizations that intentionally perform actions to alter the food characteristics or disrupt the supply chain to prevent its functionality. Food safety refers to the development of actions to reduce the likelihood of food contamination and prevent the resulting harmful consequences of unsafe food, such as illness and death (Akkerman et al., 2010; Marucheck et al., 2011). A supply chain perspective can highlight the safety problems that arise from transfers along the chain, such as improper storage, handling and distribution of the food (Marucheck et al., 2011). Because we are interested in understanding the development of studies on food contamination in the supply chain management, we focus on food safety.

3. Methods We developed a systematic literature review to analyze a sample of papers published about supply chain management issues related to food safety. The systematic literature review is considered appropriate when the principal purpose of the study is to summarize the studies on a given topic. It allows for reduction of many pieces of literature in an explicit and systematic way, providing a short coherent report that helps readers to know and understand something about a given topic (Pittaway et al., 2004; Tranfield et al., 2003). This

International Food and Agribusiness Management Review

101

Auler et al.

Volume 20, Issue 1, 2017

method helps the researcher reduce his/her research bias and be more explicit in terms of research choices (Akobeng, 2005; Tranfield et al., 2003). The first step is to choose the keywords to be used in the selection process of the papers (Tranfield et al., 2003). We chose the keywords, ‘supply chain’ and ‘food safety’, because they are common to most papers about this topic. The second step is to determine the database to be used for data collection. We chose the Web of Science database, because this is one of the most complete and cited reference databases in the supply chain management field. The third step is to determine the period for data collection. We did not set any restrictions related to time, allowing for the inclusion of papers published in any year. The fourth step is to determine the data collection categories. We chose the categories, ‘Business’, ‘Management’ and ‘Operations Research Management Science’, because our study focuses on studies related to business and management. It is important to note that we excluded studies about the production process of food, such as agriculture, chemistry, nutrition, biology or food engineering, given that we focused on the management of the supply chain after the production process is concluded and not during the production process itself. In summary, we entered the keywords, ‘supply chain’ and ‘food safety’ and the aforementioned categories into the search tool on the Web of Science database to refine our search and obtain a sample of papers to be analyzed in our study. This process was conducted in June 2015, and we found 47 papers that matched our sample criteria. The final step consists of reading the abstract of each paper to evaluate its adherence to our study. To do so, we used the HistCite software (Thomson Reuters, New York, NY, USA) to export the information about each paper to a word processor. After reading the abstracts, we excluded one paper because it was a retraction. Our final sample was composed of 46 papers (Supplementary Table S1).

4. Results Journals and citation results Table 1 presents information about the journals and their Journal Citation Report (JCR) Impact Factor 2014 Edition, the number of papers in each journal, and the citation score in the Web of Science database. The 46 papers in our sample were published in 23 journals. The JCR impact factor of these journals varies from 0.386 to 4.376. The International Journal of Production Economics published 10 papers, which represent 21.74% of the sample, and was the journal containing the greatest number of published papers about supply chain management and food safety. Another 15 journals (65.22% of the total) published only one paper. These results show a concentration of 31 papers published in 8 journals. This means that 8 journals account for 67.39% of all papers published on food safety and supply chain management in our sample. The citation score in the Web of Science database refers to the number of citations that the papers of a particular journal received in papers indexed by the Web of Science database. For example, there is only one paper published in the Journal of Supply Chain Management that received 82 citations in journals indexed by the Web of Science database. Another example is the International Journal of Production Research, which has 3 papers published about this topic that received 50 citations. Another characteristic of papers in our sample is the publication year, which provides information about the evolution of publications at the time. As we can see in Figure 2, all the papers were published after 2004. It is important to note that this research was conducted in July 2015, considering only papers published until that date. Figure 2 shows that an increase in the number of papers published about food safety. One reason for such an increase may be the growth in the number of cases of food contamination in recent years.

International Food and Agribusiness Management Review

102

Auler et al.

Volume 20, Issue 1, 2017

Table 1. Journals in the sample. Journal

1 Omega-International Journal of Management Science 2 Journal of Supply Chain Management 3 Journal of Operations Management 4 Supply Chain Management: An International Journal 5 International Journal of Production Economics 6 Technovation 7 European Journal of Operational Research 8 Expert Systems with Applications 9 International Journal of Electronic Commerce 10 International Journal of Physical Distribution and Logistics Management 11 International Journal of Production Research 12 Production Planning and Control 13 Journal of Business Ethics 14 International Journal of Consumer Studies 15 Business Horizons 16 OR Spectrum 17 International Journal of Logistics Management 18 Information Technology and Management 19 International Journal of Shipping and Transport Logistics 20 Service Industries Journal 21 Journal of Business and Industrial Marketing 22 International Journal of Logistics Research and Applications 23 Transportation Journal Total 1

Papers in the sample

Citation scores in Web of Science

4.376 3.857 3.818 3.500 2.7522 2.526 2.358 2.240 1.872 1.802 1.477 1.466 1.326 1.293 1.163 0.987 0.946 0.8972 0.862 0.8322 0.750 0.482 0.386

1 1 1 3 10 1 6 2 1 1 3 1 1 2 1 2 3 1 1 1 1 1 1 46

– 82 25 6 76 49 60 8 2 4 50 8 22 2 – 58 – 10 – – 4 7 – 473

JCR = Journal Citation Report. Impact factor in 2012 JCR edition.

Number of papers

JCR1 Impact Factor 2014

12 10 8 6 4 2 0

2004 2006 2008 2009 2010 2011 2012 2013 2014 2015* Publication year

Figure 2. Number of papers by publication year (*papers published from January to July). Authorship results Table 2 shows the number of authors in the same paper, the number of papers published, and the authorship. The number of authors in the same paper varies from 1 to 7. There are 34 papers with 2 and 3 authors, which represent 73.91% of the sample. The authorship refers to the number of papers written by a given author. This means that one author can have written multiple papers. For this reason, the third column multiplies International Food and Agribusiness Management Review

103

Auler et al.

Volume 20, Issue 1, 2017

Table 2. Papers by number of co-authors. Authors in the same paper

Published papers

1 1 2 18 3 16 4 7 5 1 6 2 7 1 Total 46 Average mean of co-authors by paper

2.98

Authorship (authors × papers)

1 36 48 28 5 12 7 137

2.17 39.13 34.78 15.22 2.17 4.35 2.17 100.00

the number of authors by the number of papers, providing information to help us create a map of citations and summarize the authorship by the country of author affiliation, which is presented later. A total of 126 authors results in 137 authorships, which means that some authors wrote more than one paper. To better understand the impact of some authors and their papers on the development of other studies about food safety and supply chain management, we developed a citation network analysis. For this analysis, we consider only papers that receive the citation by papers of our sample, that is, papers that are not in our sample are not considered for this analysis. Figure 3 illustrates the results. Each ‘arrow out’ of a box represents a citation received by the paper in that box. For example, the paper by Roth et al. (2008) was cited 8 times by other papers in the sample. Because this paper is the most cited paper, it is in the center of the network. The more centralized a paper is, the more citations it received. Thus, each circle in Figure 3 represents the number of citations that papers in that circle received. Conversely, papers in the outer circle did not receive any citation. It is important to note, however, the most recent papers tend to be in the outer circle, while

Figure 3. The citation network analysis. Each ‘arrow out’ of a box represents a citation received by the paper in that box. Each circle represents the number of citations that papers in that circle received. International Food and Agribusiness Management Review

104

Auler et al.

Volume 20, Issue 1, 2017

older papers in the center. The number of citations may be related to the time a paper is available to be cited by other authors, influencing the position of these papers in our analysis. Table 3 shows the country of the institution where the author is affiliated. There are authors from institutions of 20 countries. Most authors are from the United States and the United Kingdom, 30 and 19, respectively. This corresponds to 21.89 and 13.87% of all authors in our sample. On the other hand, Finland, India, Mexico, Norway, and Sweden have one author each. Figure 4 illustrates the country of the institution where the first author is affiliated. This figure shows the concentration of authors in developed countries located in North America and Europe. It is interesting to note that some countries that are big food producers, such as Brazil, Australia, China and India have few authors that published papers about food safety. Methodological characteristics We also analyzed some methodological characteristics of the papers. First, we examined the study design adopted by authors considering: (1) a qualitative approach; (2) a quantitative approach; and (3) a combination of these two approaches. Out of 46 papers, 30 (65.22%) are based on quantitative design, 12 (26.09%) are based on a qualitative design, and 4 (8.69%) are based on a quali-quantitative design. These results suggest a trend toward more quantitative studies. Also, although the number of papers with a quali-quantitative design is not high, it suggests that some authors employed a more sophisticated methodological design to answer their questions, as recommended by theorists like Shah and Corley (2006) and Weick (1995). Table 3. Country of the institution where the author is affiliated. Country of institution

Australia Brazil Canada China Denmark Finland France Germany Hong Kong India Italy Mexico The Netherlands Norway Saudi Arabia Spain Sweden Taiwan United Kingdom United States Total

Order of authorship 1st

2nd

3rd

4th

2 2 4 2 3 1 1 4 2 1 3

2 1 3 2 3

1 1 1 2

1 4 2

1 5 2

2 1 1

2 1 6 11 46

3 1 2 7 10 45

5th

6th

1 2

1 1 1

4 6 27

1 2 11

1 4

International Food and Agribusiness Management Review

105

7th

1 3

Total 6 4 8 10 6 1 3 14 10 1 6 1 6 1 2 5 1 3 19 30 137

Auler et al.

Volume 20, Issue 1, 2017

Figure 4. First author by country. Second, we examined if the paper is: (1) purely theoretical; (2) analytical models; and (3) theoretically driven and empirically tested. Furthermore, for the papers that are theoretically driven and empirically tested, we analyzed their data sources: primary, secondary or a combination of these two. Figure 5 shows the results. Of the 46 papers analyzed, 5 (10.87%) are purely theoretical, 14 (30.43%) are analytical models, and 27 (76.09%) are theoretically driven and empirically tested. A total of 8 analytical model papers employed empirical data to test their models, while the other 6 did not make use of any type of data. Of the 27 theoretically driven and empirically tested papers, 16 used primary data sources, 8 used secondary data, and 3 used a combination of primary and secondary data sources. Analysis of the primary data sources reveal that survey was employed in 9 papers, case study was employed in 5 papers, and interview was used in 2 papers. Regarding the major subject of each paper, traceability was the most analyzed subject, accounting for 10 papers (21.74% of the sample). In this group, the way supply chain managers could use systems and tools to track or trace raw materials and final products particularly in food chains was the central point of the discussions. The second most studied subject was risk, amounting to 5 papers (10.87%). Here, supply chain scholars explored some alternatives to overcome problems that practitioners could face when supplying raw materials and final food products, both being proactive or reactive to such problematic issues. Another major subject, distribution accounted for 5 papers (10.87%). Distribution of food products is particularly important to supply chain practitioners and scholars, given the inherent characteristics of food, such as perishability, health or even safety. The way practitioners could better perform distribution, taking into account these characteristics, was the major point in this group. Finally, 4 papers (8.70%) discussed quality as the major subject. Here, issues related to what supply chain practitioners could do to deliver food products at highquality levels, especially in accordance with international quality standards, was the major point of analysis. To summarize, these four subjects were the most important themes discussed in our sample, accounting for 24 papers (52.17%).

Purely theoretical 16

27 5

Analytical models Theoretically driven and empirically tested with primary data sources

Theoretically driven and empirically tested with secondary data sources

Theoretically driven and empirically tested with primary and secondary data

Figure 5. Type of paper and the sources of data. International Food and Agribusiness Management Review

106

Auler et al.

Volume 20, Issue 1, 2017

Results about major subjects are in agreement on the relevance of papers considering the number of citations by papers in the sample, as in the case of network citation (Figure 3). In this case, there are 14 papers that received this kind of citation, and traceability is the central point in 4 of them, risk in 3, distribution in 2, quality in 2, food safety plan, recall and stock accounted for 1 paper each. Citing the papers and considering the network citation again, the central point of relevance is the paper written by Roth et al. (2008) discussing supply chain quality management, followed by Wang et al. (2009) analyzing quality, Fritz and Schiefer (2009) discussing traceability, Wang et al. (2010) examining traceability, and Kumar and Budin (2006) studying recall. All of these results revealed the same idea, that traceability, quality, risk, and distribution are the most relevant aspects of food supply chain management in the sample. Finally, we analyzed in which countries authors focused on testing their research ideas in the papers of our sample, that is, what country, or group of countries, was/were the target of analysis. It is important to note that 33 papers (71.73%) mentioned which country/countries was/were analyzed and 13 (28.27%) cited any, because they did not analyze empirical data. In these 33 papers, there are 47 indications of target countries, since a given paper could analyze more than one target country. For instance, a given paper could analyze China, the United Kingdom, and Australia, accounting for 3 target countries and summing up only one paper. Thus, the set of 33 papers cited 2 continental regions and 19 countries as their targets. From the continental regions point of view, Europe was cited twice, and Latin America was cited once. Regarding countries, the United States was the target country of 7 papers, the United Kingdom 6, and China 5, as shown in Figure 6. Together, these three target countries accounted for 38.30% of the total number of targets in the sample, showing their importance in the studies about food safety in the supply chain. If we group the 47 targets into continental regions, Europe accounts for 21 targets (44.68%), North America for 11 (23.40%), Asia 11 (23.40%), Oceania 3 (6.38%), and Latin America only 1 (2.13%). It is worth noting that Latin America was cited only as a continental region and not its individual countries. Theoretical approaches We also analyzed the theoretical approaches in the papers. Figure 7 shows that 36 (78.26%) papers were based on concepts and models from the supply chain management perspective, such as, distribution management, risk assessment, supply chain performance, traceability, transportation planning, factors influencing supplier selection, and batch dispersion. The other 10 papers are based on theoretical approaches, such as, agency theory, institutional theory, resource-based view, among others. Figure 7 summarizes the results. Two papers adopted a marketing perspective. The first paper discusses results of traceability systems from the customersâ&#x20AC;&#x2122; point of view about sustainable products. The second paper uses concepts of attributes by the end customers to buy their food, especially beef. One paper employed the Porterâ&#x20AC;&#x2122;s five forces and a SWOT analysis to evaluate best practices for prevention and management of product recalls. Finally, three

Figure 6. Target country of the studies. International Food and Agribusiness Management Review

107

Auler et al.

Volume 20, Issue 1, 2017

Agency theory (1) Graph theory (1) Institutional theory (1) Supply chain concepts (36)

Porterâ&#x20AC;&#x2122;s five forces and SWOT (1) Fuzzy set theory (2)

Other (10)

Marketing literature (2) Resource based view (2)

Figure 7. Theoretical approaches in the sample. papers were based on general theories. The first employed graph theory to compose the network of the dairy industry in Germany, mapping the supply chain and analyzing risks of deliberate food contamination. To solve problems with a large range of uncertainty, the other two papers employed fuzzy theory as a methodological approach to help analysis of risk assessment, and supplier selection attributes in a supermarket context.

5. Discussion Because of the growing number of events of food contamination and because of their relationship with supply chain management, we are motivated to conduct a systematic literature review to describe and understand characteristics of papers published about food safety in the area of supply chain management studies. Our results show an increase in the number of papers published on this topic, which is consistent with the growing number of food contamination events in the last few years. This may also be the reason we see many journals publishing papers about it. As the food supply chain becomes more global in the sense that more food producers are spread out all over the world, food safety is an important factor in delivering food products to consumers. Some papers in our sample tend to be more influential than others. This is the case of papers in more central positions of our citation network analysis. One explanation for this position may be the topics and results presented by these papers, which create research opportunities for other authors. For instance, the Roth et al. (2008) paper discusses issues that impact food safety and presents a framework containing six constructs to improve food safety. Given its many propositions, this paper may serve as a source of ideas for other authors. Thus, depending on the objective of the paper, it may provide more ideas to another author. More theoretical papers may serve as a basis for the development of ideas that are tested by other authors. Another explanation may be the year in which the paper was published. The older the paper, the more likely it tends to be cited because more authors are likely to read it. The country of origin of the first authorâ&#x20AC;&#x2122;s institutions suggests a concentration of papers in developed countries, mostly from the United States and the United Kingdom. The higher number of institutions and researchers in these countries may help to explain these results. The country of origin of the first authorâ&#x20AC;&#x2122;s institution is important to show which countries and institutions are conducting more research about food safety and supply chain management. Other authors interested in partnering with scholars publishing about this topic may find this information useful. Also, other institutions can identify those institutions that have higher numbers of publications and may seek partnership.

International Food and Agribusiness Management Review

108

Auler et al.

Volume 20, Issue 1, 2017

Considering the target country, however, China and Australia have to be taken into account jointly with North America and Europe. Even if the group of analyzed datasets continues to pay major attention to developed countries, the emerging market China is an important player in global markets as well as in the global economy. However, some developing countries considered big food producers, such as Brazil and India, have been less investigated. Taken together, these results suggest opportunities for future research by having researchers from developed and developing countries working together to investigate ideas and collect data from their countries. Scholars from institutions located in developing countries may have more information and more knowledge about food supply chains in these countries and, therefore, may contribute with different ideas for research projects developed by their peers in developed countries. Also, collecting data from countries that are big food producers could help to reveal their food safety and supply chain management practices, providing more ideas to better understand and explain these issues. The methodology adopted by authors in the analyzed papers show the predominance of a theoretically-driven, empirically-tested approach based on the quantitative design to analyze empirical data. We note a balance between the use of primary and secondary datasets, which may enrich results about food safety and supply chain management, because it is possible to have an objective perspective by using secondary data, and a subjective perspective by using surveys or interviews. Two literature reviews are noteworthy, which suggests that some scholars are trying to collect data on what other authors are doing in an attempt to synthesize their findings. A smaller number of papers were based on analytical models. This variety of methodological approaches demonstrates that this topic has called the attention of researchers from many methodological backgrounds. These results also suggest that an effort will be needed to organize these findings and reconcile the literature on sub-topics of food safety and supply chain management. The theoretical approaches used by authors to view food safety problems are more related to the supply chain management literature. It is important to draw attention to the fact that the supply chain management literature is an evolving discipline (Harland et al., 2006) and still needs more theoretical robustness to become a theory (Burgess et al., 2006). For this reason, as the theoretical background for their ideas, most studies employed supply chain management concepts and models, such as, distribution management, risk assessment, supply chain performance, traceability, to cite just a few. The application of these concepts and models is expected given the supply chain management perspective adopted by these papers. This is also the reason few papers employed more robust theoretical approaches, such as agency and institutional theories, to explain a phenomenon that exists along the chain of goods and information. Even so, there is still room for use of more robust theories, such as, social capital theory (Burt, 1997; Coleman, 1988), transaction cost theory (Williamson, 1985), and population ecology theory (Hannan and Freeman, 1977), as well as theoretical perspectives, such as, organizational power (Hickson et al., 1971; Pfeffer, 1981) and property rights (Demsetz, 1967).

6. Conclusions The objective of our paper was to review a sample of papers published about food safety and supply chain management to understand the main characteristics of these papers. We conducted a systematic literature review using the keywords ‘food safety’ and ‘supply chain management’ in the Web of Science database, and found 46 papers that match our requirements. The results show a concentration of publications from authors of North American and European countries, using data also collected in these countries and employing quantitative methods based on analytical models and a theoretically-driven, empirically-tested approach. The theoretical approaches used by most papers come from the supply chain management literature and use concepts and models, such as, those related to traceability, transportation planning, factors influencing supplier selection, and batch dispersion. Our paper contributes to the literature on food supply chain management by showing how scholars have been working to understand and solve problems related to food safety in the supply chain arena. By doing so, we reveal the path followed by previous scholars in the process of conducting their research projects regarding International Food and Agribusiness Management Review

109

Auler et al.

Volume 20, Issue 1, 2017

methodological and theoretical issues, as well as network citations and institution nationalities where the studies were conducted. Future scholars can follow a similar path to contribute to the research about the topic discussed in our paper, as well as contact partners in accordance with their own research topic(s). In addition, other scholars can perceive research opportunities by following a different path, designing their studies using distinct methodological or theoretical approaches, rather than those included in our paper. Both paths could provide new insights into the food safety supply chain phenomena. One limitation of our paper relates to the keywords and database used to collect the papers in our literature review. For example, we did not include the keyword ‘food contamination’, which is a keyword somewhat related to ‘food safety’ and could increase the number of papers in our sample. Other studies should include other related keywords to expand the search for papers. In addition, other studies could include other databases to capture a higher number of papers. Another research opportunity is to include a higher number of keywords and databases, but restrict the journals to be analyzed based on some criteria, such as, the journal impact factor and the ‘H’ index. Finally, other studies could evaluate the number of papers published about food safety and supply chain management relative to the total number of papers published by the journal to provide more accurate information about the importance of this topic to a given journal.

Supplementary material Supplementary material can be found online at https://doi.org/10.22434/IFAMR2016.0003. Table S1. Analyzed papers.

References Akkerman, R., P. Farahani and M. Grunow. 2010. Quality, safety and sustainability in food distribution: a review of quantitative operations management approaches and challenges. OR Spectrum 32, 863-904. Akobeng, A.K. 2005. Understanding systematic reviews and meta-analysis. Archives of Disease in Childhood 90: 845-848. Alfaro, J.A., and L.A. Rábade. 2009. Traceability as a strategic tool to improve inventory management: a case study in the food industry. International Journal of Production Economics 118: 104-10. Beske, P., A. Land and S. Seuring. 2014. Sustainable supply chain management practices and dynamic capabilities in the food industry: a critical analysis of the literature. International Journal of Production Economics 152: 131-343. Burgess, K., P.J. Singh and R. Koroglu. 2006. Supply chain management: a structured literature review and implications for future research. International Journal of Operations and Production Management 26: 703-29. Burt, R.S. 1997. The contingent value of social capital. Administrative Science Quarterly 42: 339-365. Chebolu-Subramanian, V. and G.M. Gaukler. 2015. Product contamination in a multi-stage food supply chain. European Journal of Operational Research 244: 164-175. Coleman, J.S. 1988. Social capital in the creation of human capital. American Journal of Sociology 94 (Supplement): S95-120. Dani, S. and A. Deep. 2010. Fragile food supply chains: reacting to risks. International Journal of Logistics Research and Applications 13: 395-410. Demsetz, H. 1967. Toward a theory of property rights. American Economic Review 57: 347-59. Ding, M.J., F. Jie, K.A. Parton and M.J. Matanda. 2014. Relationships between quality of information sharing and supply chain food quality in the Australian beef processing industry. International Journal of Logistics Management 25: 85-108. Epelbaum, F.M.B. and M.G. Martinez. 2014. The technological evolution of food traceability systems and their impact on firm sustainable performance: a RBV approach. International Journal of Production Economics 150: 215-224.

International Food and Agribusiness Management Review

110

Auler et al.

Volume 20, Issue 1, 2017

European Food Safety Authority (EFSA). 2008. Statement of EFSA on the risks for public health due to the presence of dioxins in pork from Ireland.’ EFSA Journal 911: 1-15. Fritz, M. and G. Schiefer. 2009. Tracking, tracing, and business process interests in food commodities: a multi-level decision complexity. International Journal of Production Economics 11: 317-329. Hannan, M.T. and J. Freeman. 1977. The population ecology of organizations. American Journal of Sociology 82: 929-964. Harland, C.M., R.C. Lamming, H. Walker, W.E. Phillips, N.D. Caldwell, T.E. Johnsen, L.A. Knight and J. Zheng. 2006. Supply management: is it a discipline? International Journal of Operations and Production Management 26: 730-753. Hickson, D.J., C.R. Hinings, C.A. Lee, R.E. Schneck and J.M. Pennings. 1971. A strategic contingencies’ theory of intraorganizational power. Administrative Science Quarterly 16: 216-229. Kumar, S. and E.M. Budin. 2006. Prevention and management of product recalls in the processed food industry: a case study based on an exporter’s perspective. Technovation 26: 739-750. Lao, S.I., K.L. Choy, G.T.S. Ho, Y.C. Tsim, T.C. Poon and C.K. Cheng. 2012. A real-time food safety management system for receiving operations in distribution centers. Expert Systems with Applications 39: 2532-2548. Layton, L. and N. Miroff. 2009. The rise and fall of a peanut empire. The Washington Post, February 15, 2009. Marucheck, A.S., N. Greis, C. Mena and L. Cai. 2011. Product safety and security in the global supply chain: issues, challenges and research opportunities – editorial essay. Journal of Operations Management 29: 707-720. Pérez-Mesa, J.C. and E. Galdeano-Gómez. 2015. Collaborative firms managing perishable products in a complex supply network: an empirical analysis of performance. Supply Chain Management 20: 128-138. Pfeffer, J. 1981. Power in organizations. Pitman, Boston, MA, USA. Pittaway, L., M. Robertson, K. Munir, D. Denyer and A. Neely. 2004. Networking and innovation: a systematic review of the evidence. International Journal of Management Reviews 5-6: 137-168. Reiner, G. and M. Trcka. 2004. Customized supply chain design: problems and alternatives for a production company in the food industry. a simulation based analysis. International Journal of Production Economics 89: 217-229. Resende-Filho, M.A and T.M. Hurley. 2012. Information asymmetry and traceability incentives for food safety. International Journal of Production Economics 139: 596-603. Rong, A. and M. Grunow. 2010. A methodology for controlling dispersion in food production and distribution. OR Spectrum 32: 957-978. Roth, A.V., A.A. Tsay, M.E. Pullman and J.V. Gray. 2008. Unraveling the food supply chain: strategic insights from china and the 2007 recalls. Journal of Supply Chain Management 44: 22-39. Shah, S.K. and K.G. Corley. 2006. Building better theory by bridging the quantitative – qualitative divide. Journal of Management Studies 43:1821-1835. Sloane, A. and S. O’Reilly. 2013. The emergence of supply network ecosystems: a social network analysis perspective. Production Planning and Control 24: 621-639. Spencer, R. 2009. Two sentenced to death over china melanine milk scandal. The Telegraph, January 22, 2009. Available at: http://tinyurl.com/b722s6. Tang, C.S. and V. Babich. 2014. Using social and economic incentives to discourage Chinese suppliers from product adulteration. Business Horizons 57: 497-508. Ting, S.L, Y.K. Tse, G.T.S. Ho, S.H. Chung and G. Pang. 2014. Mining logistics data to assure the quality in a sustainable food supply chain: a case in the red wine industry. International Journal of Production Economics 152: 200-209. Tranfield, D., D. Denyer and P. Smart. 2003. Towards a methodology for developing evidence-informed management knowledge by means of systematic review. British Journal of Management 14: 207-222. Van der Gaag, M.A., F. Vos, H.W. Saatkamp, M. Van Boven, P. Van Beek and R.B.M. Huirne. 2004. A statetransition simulation model for the spread of Salmonella in the pork supply chain. European Journal of Operational Research 156: 782-798.

International Food and Agribusiness Management Review

111

Auler et al.

Volume 20, Issue 1, 2017

Wang, X., D. Li and C. Oâ&#x20AC;&#x2122;Brien. 2009. Optimisation of Traceability and operations planning: an integrated model for perishable food production. International Journal of Production Research 47: 2865-2886. Wang, X., D. Li, C. Oâ&#x20AC;&#x2122;Brien and Y. Li. 2010. A production planning model to reduce risk and improve operations management. International Journal of Production Economics 124: 463-474. Wang, L., H.-B. Shi, S. Yu, H. Li, L. Liu, Z. Bi and L. Fu. 2012. An application of enterprise systems in quality management of products. Information Technology and Management 13: 389-402. Weick, K.E. 1995. What theory is not, theorizing is. Administrative Science Quarterly 40: 385-390. Williamson, O.E. 1985. The Economic Institutions of Capitalism. The Free Press, New York, NY, USA.

International Food and Agribusiness Management Review

112

OPEN ACCESS International Food and Agribusiness Management Review Volume 20 Issue 1, 2017; DOI: 10.22434/IFAMR2015.0051 Received: 28 April 2015 / Accepted: 18 October 2016

Food scare crisis: the effect on Serbian dairy market RESEARCH ARTICLE Rade Popovic a, Boris Radovanovb, and James W. Dunnc aAssociate

professor, Department for agricultural economics and agribusiness, Faculty of Economics Subotica, University of Novi Sad, Segedinski put 9-11, 24000 Subotica, Serbia

bAssistant

professor, Department for business informatics and quantitative methods, Faculty of Economics Subotica, University of Novi Sad, Segedinski put 9-11, 24000 Subotica, Serbia cProfessor,

Department for agricultural economics, sociology and education, College of agricultural sciences, Penn State University, 203 Armsby, University Park, PA 16802, USA

Abstract The increasing trend of food scandal crises is not well followed in recent studies of spatial price transmission. This paper analyses the impact on the domestic market of an Aflatoxin M1 outbreak in the Serbian dairy sector during 2013/2014 using a spatial price transmission approach. Monthly farm milk prices in Serbia for the period 2007/2014 were contrasted with leading dairy exporting countries New Zealand, USA and Germany, which did not have a food scare in their dairy sectors. To estimate the impacts a Markov-switching vector error-correction model was utilized. For all four dairy markets the model identified two price change regimes: standard and extreme. Although it was predictable, an extreme regime was not identified during the Aflatoxin M1 crises in Serbia because of some specific characteristics of its dairy production. The results suggest that the Aflatoxin M1 outbreak ‘froze’ the Serbian dairy market and temporally disconnected it from the world milk market. Farmer’s prices fell below their long-run equilibrium levels. The total loss of the Serbian farm-level dairy sector during the crisis reached up to 96.2 million EUR. These ‘missed opportunity’ significantly slowed investment in the dairy sector. Keywords: food crisis, aflatoxin M1, milk market, spatial price transmission, Serbia JEL code: Q17, Q18 Corresponding author: popovicr@ef.uns.ac.rs

113

Popovic et al.

Volume 20, Issue 1, 2017

1. Introduction Milk has a special place in people’s diets, and is an especially important food for children. As such, whenever an issue of milk quality arises, the consumer response is rapid and serious. In early 2013, Serbia had an Aflatoxin M1 outbreak that led to product recalls and a dramatic decrease in purchases. This paper examines this outbreak and its impact on Serbian markets. Outcomes from this study provide several lessons learned that can be used for better reaction from management in the food sector and government during future food safety crises. The purpose of this paper is to quantify the effect of the Aflatoxin M1 crisis on the Serbian dairy market using a spatial price transmission model. The objectives of the paper are: to measure the effect of this crisis on the domestic milk market and to determine changes in the degree of Serbian milk market integration in the world milk market. When a food scandal arises, three parties are under pressure. The government faces political responsibility for legislative regulation of food safety standards (Bergeaud-Blackler and Forretti, 2006). Consumers faced with a safety incident stop consuming that type of food from one company, or altogether if incidence is not limited to a particular firm (Mitchell, 2003; Banati, 2011), as was the case in Serbia. Food companies suffer losses from: food product recalls from shelves, a drop of sales on domestic and international markets, lost consumer confidence, etc. With globalisation of food production and trade, the consequences of food scandals usually extend beyond the domestic market. Often today the strategic orientation of food companies on the international market additionally increases the possibilities of negative consequences. Maize produced in the autumn of 2012 was identified as a main source of the Aflatoxin in milk. Hot and dry weather conditions during the second half of the maize production season produced mould and toxin production. Besides that, poor postharvest handling, storage and manufacturing practices can influence mould growth and toxin production (Hussaini, 2013; Skrbic et al., 2014). In the past occurrences of Aflatoxin in maize production in Serbia were rare. The presence of Aflatoxin in maize throughout the main production areas in Serbia for the period 2009-2012 was observed only in 2012, in 68.5% of the samples (Kos et al., 2013). The toxin concentration in 29.5% of the samples was very high (>50 μg/kg), and, according to current standards, such maize couldn’t be used for food or livestock feed. Regarding climate changes, Serbia may become more vulnerable to temperature problems that could create aflatoxins in the future (Kos et al., 2013). Today, most countries with higher food quality standards face occasional food scares and scandals. When a food scandal arises it has a negative effect on the whole food supply chain, but especially on its most vulnerable link, the farmers. The literature analysing the effect on the farm level predominantly focuses on vertical price transmission on national markets (Hasssouneh et al., 2010, 2012; Lloyd et al., 2001; Popovic and Radovanov, 2010). Only in some exceptional cases was a different approach used. Recently, researching the effect of export controls, spatial price transmission was applied to quantify the effect on farm wheat prices in exporting countries during a food crisis (Djuric et al., 2012; Gotz et al., 2013). There are several reasons why analysis of the Aflatoxin M1 outbreak effects on the Serbian dairy market is important. First, the crisis lasted almost two years, and the magnitude of the crisis was significant. Second, there wasn’t a single dairy company source of the crisis, but rather all dairy companies were involved. Third, the economic and social importance of the dairy sector in Serbia is high. Milk production accounts for 7% of agriculture output. From the social side, one fourth of the 632 thousand farms in Serbia are dairy farms. Among agricultural policy measures, milk premiums are the most important single coupled measure of direct payment. Also, Serbia is a net exporter of dairy products. The average self-sufficiency, calculated in milk equivalents (ME), in dairy products during period 2005-2014 was 102.5%. Calculation of ME is based on fat and protein content in dairy products (Hemme, 2008). Besides that, the crisis cut investments in the dairy sector, which are essential to prepare the Serbian agricultural sector for potential EU membership.

International Food and Agribusiness Management Review

114

Popovic et al.

Volume 20, Issue 1, 2017

The paper is organized in five related sections. After the introduction, the first section covers the literature review of food scares and scandals and the econometric methodology used for this topic. The second section explains the milk aflatoxin crisis in Serbia and how it affected milk price development. The third section describes data sets and methodology used to capture the effect on the price transmission. The fourth section outlines the main results from the Markov switching vector error correction model (MS-VECM) specification. At the end, the fifth section of this paper offers some recommendations and concluding remarks.

2. Literature review Studies of food scares and scandals are not numerous and most analyse developed markets. The majority focus on effects along the food supply chain. Vertical price transmission is the main approach in the analysis of a crisis and its effect on domestic markets. The most common finding is that different levels of the food supply chain respond differently to market shocks. During the bovine spongiform encephalopathy (BSE) outbreak in the UK beef market, retail prices declined, but less than farm prices (Lloyd et al., 2001). Similar conclusions were reached in investigation of the impact of a BSE outbreak on the Spanish beef market, and later on the impact of avian influenza on the Egyptian poultry market (Hassouneh et al., 2010, 2012). Studies of vertical price transmission in Serbian dairy market (Popovic et al., 2013; Popovic and Radovanov, 2010) confirm also that information asymmetry exists and world milk price signals do not pass quickly and equally through subsequent levels of the Serbian milk supply chain. The approach applied differs from previous studies in the use of a spatial price transmission model approach to identify losses on the Serbian dairy market, because of missed higher price levels on the world market during a two year period. It is assumed that the milk aflatoxin crisis temporally changed the price relationships between the domestic and the world market. This paper hypothesizes that the food scandal crises (milk aflatoxin M1) decreased the level of long-run price transmission and created negative effects on dairy farm economics. Early empirical findings of price transmission were based on simple correlation and regression analysis that did not involve dynamics and lagged variables in detecting price relationships (Fackler and Goodwin, 2001). The emerging co-integration studies emphasized a few drawbacks related with the price regression analysis. Specifically, regression can lead to spurious results when price data are non-stationary (Hassouneh et al., 2010). The first study of price transmission on agricultural markets using co-integration methods was published in 1989 (Ardeni, 1989). Today almost the entire empirical price transmission literature employs co-integration methods, especially vector error-correction models (VECM) (Barrett and Li, 2002). For instance, a threshold vector error-correction and a threshold autoregression model capture non-linearities and were first introduced in spatial price transmission analysis (Goodwin and Piggott, 2001). Also, evidence of threshold behaviour in cases with differences between thresholds for the wheat and maize prices were found in price transmission between wheat markets in: Brazil, Argentina and USA (Balcombe et al., 2007). On the other hand, a smooth transition vector error correction model, developed by Terasvitra, does not presume that regime shifts are sudden, but instead of that regimes shift regularly (Terasvitra, 1994). Finally, a Markov switching vector error correction model was introduced in an analysis of business cycles and with one obvious distinguishing feature that the regime switches are driven by a probabilistic variable, whereas all of the above model specifications assume the regime switches deterministically (Krolzig et al., 1997). The same model was used for investigation of the effects of an unstable policy environment on vertical price transmission (Brummer et al., 2009). Recently the mentioned model was used to study the influence of export restrictions and domestic market policy changes on horizontal price transmission in the case of wheat in Russia and Ukraine during the 2007/2008 global food crisis (Gotz et al., 2013). A comparable approach is applied to analyse the effects of governmental market interventions during the commodity price peaks on the transmission of price changes along the wheat-to-bread supply chain in Serbia (Djuric et al., 2012).

International Food and Agribusiness Management Review

115

Popovic et al.

Volume 20, Issue 1, 2017

3. Milk aflatoxin crisis and price development in Serbia Among several recent foods scare crises in Serbia (trichinosis in pork, pesticide residues in apples, ochratoxin in strudel, etc.) none attracted the attention of the public and politicians to the degree of the 2013 aflatoxin M1contamination of milk. Also it lasted the longest, almost two years, from February 2013 through 2014. The main effects of the crisis could be analysed in two aspects: political measures and the milk supply chain adjustment. Information about high aflatoxin contamination of corn in Serbia was first published in December 2012. Initially, authorities denied the outbreak’s existence, treating the information as malevolent. The consumer scare and political adjustment to the new situation started in February 2013, when the first laboratory results of milk products were published. By the end of February, inspectors ordered the first recall of 50 types of sterilised and pasteurised milk from various milk processing companies. A few days later, the first reaction of the Ministry of Agriculture was to change the allowed level of aflatoxin in milk from 0.05 to 0.5 μg/kg, as it was until 2011. This decision generated a huge argument among politicians, experts and consumers in all kind of media. Diametrically opposite statements disturbed consumers and created a fear of milk consumption. Later results of milk sample analysis during 2013 (Kos et al., 2014; Skrbic et al., 2014) implied that the fear was justified, since all age categories, especially children, were exposed to high risk related to the presence of aflatoxin M1 in milk. In June 2013 Serbia got its first rule book about the organisation of a rapid alert system for food and feed. In the same month, after long lasting public pressure, the Minister of Agriculture resigned. This was expected, since it happened in some other countries during food scare crises (Motarjemi and Lelieveld, 2014). The new Minister announced the return to the previous standards regarding the maximum level of aflatoxin in milk, the same level used in the EU, for April 2014. The new regulations for the allowable level of aflatoxin in feed were published in March 2014 with the intention to enable milk production with lesser aflatoxin contaminant. But, since most of the 156,000 farms (RZS, 2013) produce feed on their farm, it was impossible to control and apply the new regulation. A third Minister of Agriculture, after a new government takeover in April 2014, had to deal with same crisis. On July 1, 2014 the standard of allowed contamination of milk with aflatoxin returned to the EU standard. This new standard was untenable because contaminated maize from the 2012 crop season was still in use. On July 17, standards changed from 0.05 to 0.25 μg/kg and the Ministry defended this decision with a still higher level of aflatoxin. Finally, the milk crisis from aflatoxin ended in January 2015, when the Ministry of Agriculture tightened the maximum standard for allowed level of aflatoxin in milk from 0.25 to 0.05 μg/ kg, which is the EU standard. Several mistakes occurred in the management of this crisis. Some include: parallel existence of several contradictory sources of communication, the low speed of information, a slow recall of contaminated products, and a lack of transparency and trustworthy scientific facts. The milk supply chain was strongly shaken by the aflatoxin crisis. Scared and confused consumers reacted by decreasing consumption of milk and milk products. The same reaction has been seen in many other food scandal situations (Banati, 2011.) Milk is one of most important food groups in the Serbian diet. Average consumption slowly declined, reaching 207.2 kg ME per capita in 2013. Liquid milk has biggest share of total milk consumption in Serbia. Data for 2013 reveals that consumption of pasteurized and sterilized milk in Serbia decreased by 11.4% compared to 2012, but in Belgrade it fell by 26.6% (RZS, 2014b). Also, exports of milk products to countries with higher food standards dropped suddenly. Decreased demand, both domestic and international, led to accumulation of stocks at dairy processing facilities. It increased nervousness about the financial prospects of the dairy companies. When storage facilities became full, processors started to organise sales with retailers. During a few weeks each month dairy product prices were cut by 15 to 20% from regular prices. Such sales continued almost throughout the period of crisis. It should be noted that cooperatives, which are the leading organizational form for dairy farmers in New Zealand, Germany, USA, and most other countries, don’t exist in the Serbian dairy sector. Nevertheless the Serbian dairy sector has good vertical coordination between farmers and processing companies, especially the bigger ones. During the previous decade dairy processors initiated cooperation with farmers by investing and working together to improve the quantity and quality of farm-milk production. When the milk crisis became

International Food and Agribusiness Management Review

116

Popovic et al.

Volume 20, Issue 1, 2017

obvious, processors worked with farmers, informing them how to decrease the level of aflatoxin in milk. At the beginning, they even helped farmers by offering a supply of mycotoxin absorbers as feed additives. Figure 1 illustrates the development of yearly milk net export quantities, calculated in ME, monthly farm gate milk prices for Serbia and an International Farm Comparison Network – Dairy Research Centre (IFCN) world milk price indicator in the period January 2007 to December 2014. World milk price is recalculated from USD to the national currencies for Serbia, Germany and New Zealand. Different milk products in international trade for Serbia are converted into ME by counting only the fat and protein content. The shaded field in the graph for Serbia covers the period of the aflatoxin crisis. The price data for Germany, USA and New Zealand, as big net exporters, serve for comparison for markets without a crisis in the observed period. It should be mentioned that all four markets have different dairy policies, which vary from a subsidised dairy sector in EU countries to unsubsidised in New Zealand (Hemme, 2014). Farm gate milk prices in Serbia followed world milk prices up, with significant time lags observed in 2007 and 2010. The difference between the world milk price and the Serbian farm price narrows until February 2013. This was the result of opening the Serbian market thru implementation of two trade agreements with central European countries (CEFTA) from 2007 and with the European Union, gradually from 2009 to 2014. When the aflatoxin crisis started it ‘froze’ Serbian milk prices at almost the same level. Meanwhile, the world milk price rose significantly from March 2013 to May 2014, exceeding 45 USD per 100 kg energy-corrected milk (ECM). Prices calculated in EUR (Figure 2) show even a slightly negative trend for milk prices in Serbia during the crisis. Therefore, it can be assumed that dairy farms in Serbia suffered a significant market loss during the crisis. In contrast, milk prices in Germany and New Zealand follow the world milk price, with fewer time lags and with smaller price differences. The exception is the USA during 2014. An increase in domestic milk consumption together with an increase of dairy exports and a strengthening of the USD over EUR kept USA national milk prices significantly above world milk prices.

6 4 2

Export (×106 t ME)

USA

Price (USD/100 kg ECM)

Export (×106 t ME)

nJa

70 60 50 40 30 20 10 0

New Zealand

80 70 60 50 40 30 20 10 0

8 7 6 5 4 3 2 1 0

Germany

45 40 35 30 25 20 15 10 5 0

Farm gate price

100 90 80 70 60 50 40 30 20 10 0

Price (EUR/100 kg ECM)

World market price

Export (×103 t ME)

Aflatoxin crisis

Serbia

Price (NZD/100 kg ECM)

Price (×1000 RSD/100 kg ECM)

Net export quantity

Figure 1. Development of world milk market price, farm gate milk prices, net export of milk for Serbia, Germany, New Zealand and USA (adapted from FAOSTAT, 2015; IFCN, 2015; RZS, 2014a) ECM = energycorrected milk; ME = milk equivalent.

International Food and Agribusiness Management Review

117

Popovic et al.

Volume 20, Issue 1, 2017

30 20 10 0

Ja Price (EUR/100 kg ECM) n07 Ja n08 Ja n09 Ja n10 Ja n11 Ja n12 Ja n13 Ja n14

New Zealand

40 30 20 10 0

Price (EUR/100 kg ECM)

Serbia 40

Farm gate price Germany

40 30 20 10 0 50

USA

Price (EUR/100 kg ECM) Ja n07 Ja n08 Ja n09 Ja n10 Ja n11 Ja n12 Ja n13 Ja n14

Price (EUR/100 kg ECM)

World market price (IFCN)

40 30 20 10 0

Figure. 2 Milk price pairs analysed for price transmission (adapted from IFCN, 2015; OANDA, 2015; RZS, 2014a). ECM = energy-corrected milk. Net exports of milk from Serbia in 2013 increased mainly because of a strong decline in imports, compared to exports. Higher milk prices in the region, and the world market depressed imports. During the crisis Serbia lost its position in its main export market Montenegro. Exports of liquid milk products, where perception of consumers about health issues effects was stronger, were halved. At same time, the crisis had little impact on cheese and other non-liquid milk products exports. On other side, Serbian milk products become price competitive in markets with higher aflatoxin standards up to 0.5 Îźg/kg. Exports increased mostly to the Russian market, reaching 20 million EUR. The EU sanctions to Russia in 2014 increased Russian demand for milk products from Serbia, pushing export to 31 million EUR. In same period, net exports for the main exporting countries of New Zealand, USA and Germany showed a stable increase.

4. Methodological approach and data The analysis is modelled using a non-linear price transmission approach to capture the effect on the price relationships. This paper uses a MS-VECM in order to analyse price transmission. Related to the Markov switching vector autoregression framework, a MS-VECM is a special type of the more general regimeswitching model, which can be employed in the analysis of price transmission when a few price regimes govern the market conditions and changes in the market regimes are unknown or driven by many shifts in food market policy (Hamilton, 1989). Nevertheless, market participants can alter their behaviour according to their expectations before the new price regime is introduced or overthrown. Therefore, a MS-VECM permits the recognition of different price transmission regimes, even if the state variable cannot, or can only partly be observed. The state variable indicates the predominant price transmission regime at a given point in time. Since the mentioned variable is unobserved, a stochastic process in case of the state variable is assumed. In other words, the change in regime should be considered as a random and unpredictable event. However, these price regime shifts must be involved in the investigation of the stochastic properties of the market volatility. Thus, the MS-VECM assumes that the state variable is generated from a Markov process, which is regulated by a constant transition probability matrix, where the state of the market tomorrow is driven only by the state of the market today. This model was introduced by Krolzig (1997) in order to analyse business cycles.

International Food and Agribusiness Management Review

118

Popovic et al.

Volume 20, Issue 1, 2017

Recently, Brummer et al. (2009), followed by Gotz et al. (2013), introduced this model for the analysis of price transmission. The unrestricted Markov switching vector error correction model is employed as a framework for the horizontal price transmission analysis and estimation of co-integrated relationships with the first-differenced variables and the error correction term, similar to Gotz et al. (2013): p-1

Δpt = v (st) + α (st) (β(st)' pt-1) + ∑ Ai (st) Δpt-i + εt (1) i=1

where pt indicates a vector (M × 1) of milk prices, v(st) is a vector of intercept terms or an observable regime indicator variable, α(st) is the vector of the speed of adjustment coefficients, β(st) represents the long-run co-integrating vector with one period lags and Ai are matrices (M × r) of the short-run parameters of the system that capture the autoregressive part of the price movements (M is the number of variables, r is the number of parameters). Finally, εt denotes the error term, which we assume has a zero mean and constant variance, εt ~ i.i.d.(0,∑ (st)). However, the variance can vary between regimes. The core segment of this model specification is the state variable st = 1, ..., M. This is an unobserved variable, representing which of the M possible regimes governs the model at time t. All terms in equation 1 indicate the dependence of these parameters on the state variable. The probability of being in state s in period t might depend on the full time series of variables, but the simplifying Markov assumption can be written as: P(st | st-i, Δpt-1, β'pt-1) = P (st | st-i, П),

(2)

where the square matrix Π includes the transition probabilities πij for switching from the regime in row i to the regime in column j, conditioned only on the regime in the previous period. The MS-VECM acts as an error correction mechanism in each disequilibrium regime because the regimes are generated by a stationary Markov chain (Phoong et al., 2014). The estimation of a MS-VECM is based on maximizing the likelihood function with the expectationmaximization algorithm (Krolzig, 1997). The parameters characterizing the unobserved state variables and transition probabilities are first estimated, based on estimated starting values for the parameters. In the second step, the starting values are updated from the parameters estimated in the first step within an alternative procedure. The procedure continues until estimated parameters of two consecutive estimations do not differ significantly. Data sets representing 96 monthly observations for the world market milk price and national farm-gate milk prices from January 2007 to December 2014 for Serbia, Germany, New Zealand and USA were used (Figure 2). The source of farm gate milk prices are national Statistics offices and the IFCN Database. Data for the world milk price are calculated by IFCN1. The combined IFCN world milk price indicator represents the milk price a processor could theoretically pay to its farmers, if it was selling its products on the world spot market and producing at standardised costs. It is based on weighted average of three IFCN world milk indicators: skim milk powder & butter, cheese & whey, and whole milk powder according to shares of the related commodities traded on the world market (IFCN, 2015). National and world milk prices are converted in ECM with 4% fat and 3.3% protein.

5. Empirical results At the beginning of the model estimation, the augmented Dickey-Fuller (ADF) stationarity test and the Johansen co-integration test are conducted. The ADF test with constant and linear trend, presented in Table 1, indicates that all five original data series are I(1) or non-stationary processes. At the same time, first leg differences are stationary. 1 IFCN performs analysis of dairy farm economics since 2000, reaching 98% of the total world milk production in 2014 (http://tinyurl.com/z4e6p4x).

International Food and Agribusiness Management Review

119

Popovic et al.

Volume 20, Issue 1, 2017

Table 1. Augmented Dickey-Fuller (ADF) stationarity test. Series

ADF test

Statistical significance

ptrs Δptrs ptger Δptger ptnz Δptnz ptusa Δptusa ptwr Δptwr

-2.14147 -4.74039 -2.39969 -5.22095 -2.45369 -9.68220 -1.39165 -6.36572 -2.00052 -6.12006

0.2292 0.0012 0.1446 0.0002 0.1302 0 0.5832 0 0.2862 0

The results from Table 1 give the opportunity to show that individual I(1) time series could form a stationary linear combination. The Johansen co-integration test in Table 2 reveals five co-integration equations at the 1% level of significance. In other words, the test confirms the hypothesis of existing statistically significant long-run connections among observed time series. In the next phase it is necessary to confirm the differences in the log-likelihood function values between a suggested Markov switching model and the linear VECM. The results of the LR linearity test explain the advantage of the non-linear Markov switching model over the linear vector error correction model for all four models at the 5% significance level. The Markov switching vector error correction model is estimated with varying number of regimes and lags, where the optimal model specification is chosen using the Akaike and Schwarz information criteria. The model specification consists of two different states of market integration for all four markets. The first state explains the ‘standard’ regime with a modest level of price volatility. The second state is presented as the ‘extreme’ regime with a high level of price volatility. Table 3 demonstrates the filtered regime transition probabilities that vary between 0 and 1. Therefore, Table 3 indicates the probability that one market regime switches to another. Milk price regime classification in Figure 3, shows probabilities of extreme regime for analysed markets, acquired by the model. Analysis revealed that domestic milk market in Serbia had two switches to the extreme price regime before the Aflatoxin crises. It should be noticed that gap between lower Serbian and higher world milk prices was narrowing from 2010 to January 2013. During whole period of crises, milk prices remained in the standard regime. This finding is opposite to results for some other agricultural products, for example wheat, in a period of crisis caused by rigid agricultural policy measures (Djuric et al., 2012; Gotz et al., 2013). Domestic grain markets temporally disconnected from international markets, increasing price volatility. Characteristics of dairy production could be a reason why the dairy market in Serbia continues with the standard regime during crisis. Dairy production is unique because of very long biological lags. Table 2. Johansen co-integration test. No. of CE(s)

Eigenvalue

Trace statistic

Critical value

Prob.

None At most 1 At most 2 At most 3 At most 4

0.45779 0.40479 0.25521 0.19996 0.11416

162.8347 106.5205 58.78667 31.67776 11.15295

69.8188 47.8561 29.7970 15.4947 3.84146

0 0 0 0.0001 0.0008

International Food and Agribusiness Management Review

120

Popovic et al.

Volume 20, Issue 1, 2017

Table 3. Transition probabilities.

10 0

Regime probability

nJa

Regime probability

14 n-

13 n-

nJa

USA

Regime probability

New Zealand 40

Germany 40

Regime probability

Farm gate price

World market price

Serbia 40

Price (EUR/100 kg ECM)

Regime extreme

USA

0.13088 0.98970 0.05191 0.96377 0.02655 0.89716 0.44883 0.45055

New Zealand

0.86911 0.01029 0.94808 0.03622 0.97344 0.10283 0.55116 0.54945

Germany

extreme

standard extreme standard extreme standard extreme standard extreme

standard

Price (EUR/100 kg ECM)

Serbia

To regime

Price (EUR/100 kg ECM)

From regime

Figure 3. Milk price regime classification (adapted from: IFCN, 2015; OANDA, 2015; RZS, 2014a). ECM = energy-corrected milk. Such lags cause production cycles from 10 to 15 years (Anderson et al., 2009; Ferris, 1997). In other words, once the number of cows decreased strongly, it takes many years to increase to the previous level. The milk industry in Serbia, faced with shrinking demand for dairy products during crisis, reacts in a responsible way to protect its own long-term interest, securing its input market. Dairy companies in vast number didnâ&#x20AC;&#x2122;t cut procurement of raw milk from farmers, although it was expected taking in account their short-run interest. Aware of the negative consequences for its future production, the supply of raw milk remained stable. The price of this decision for dairy companies was greatly increased stocks of dairy products, lower margins and lost profit. It inferred that the burden of crises was shared among dairy companies and farms. Additionally, feed prices faced a strong negative trend, enabling farmers to cut the cost of milk production. At the end of the Aflatoxin crises in Serbia, world milk prices fell and equalised with milk prices in Serbia. Compared to the other three markets, Serbian milk price transmission regimes are most similar to the German market. The model demonstrates that the German market was dominantly in a standard regime with several International Food and Agribusiness Management Review

121

Popovic et al.

Volume 20, Issue 1, 2017

low probabilities for switches to extreme regimes. In the period from 2013 to the end of 2014 the German dairy market was in its standard regime and prices followed the trend of world market prices. The stable domestic market was identified for New Zealand, with only two year-long periods of extreme price regimes. During period from June 2013 to June 2014 the milk market in New Zealand was in an extreme regime with volatile prices. Since milk prices in its national currency at same period were fixed, the source of price volatility in the model comes from variability of exchange rates between the NZD and the EUR. As a biggest single world milk exporter, New Zealand has significant influence on world market prices. USA has high number of switches between regimes. It demonstrates high sensitivity and fast reactions of domestic dairy market players to world market price signals. After a relatively stable period from March 2011 to the end of 2012, the milk price became more volatile again. USA milk exports increased and reinforced the USD value over the EUR, which pushed milk prices to a high value, where they remained even through 2014. The results presented in Table 4 show that there are no significant changes between the regimes for Serbia, Germany, and New Zealand. Hence, those markets retain a relatively high level of persistence for the standard regimes. The standard regime of 83 observations is detected for the Serbian market, which takes place throughout the entire observed sample period. A similar situation is noticed for Germany and New Zealand, where the standard regime remains dominant in 88 and 71 observations, respectively. The US market depicts many regime switches and the standard regime is identified in only 50 sample observations. The principal analysis of the Markov switching model parameters contains results of market integration, equilibrium, stability and duration of market effects. According to GĂśtz et al. (2013), the level of integration of the domestic markets to the world market is described by the parameters of the long-run equilibrium, the contemporaneous price transmission, and the speed of adjustment. The level of statistical significance of these parameters is estimated by the delta method (Greene, 2003). The long-run equilibrium intercept Î˛0, Table 4. Main model estimates of the MS(V)ECM.1,2

MS-VECM specification LR linearity test Probability N Integration Long run equilibrium intercept Long run slope Contemporaneous price transmission Speed of adjustment Equilibrium Average ECT Stability Standard errors Duration

Serbia

Germany

New Zealand

USA

standard extreme

MS(2)VECM(1)3 7.857 0.02 83 12

MS(2)ECM(2) 8.631 0.013 88 7

MS(2)ECM(2) 13.581 0.001 71 24

MS(2)ECM(2) 7.46 0.024 50 45

-0.024 (0.834) 1.945 (0.062) 2.339 (0.310) -0.405 (0)

3.858 (0) 2.304 (0.384) 4.715 (0)

-0.169 (0) 0.663 (0.086) 3.281 (0) -0.262 (0.001)

2.969 (0) 0.289 (0.026) 8.382 (0.265)

-0.216 (0) 0.495 (0.095) 3.789 (0) 0.894 (0)

11.759 (0) 1.078 (0.074) 22.711 (0.002)

0.596 (0.008) 1.531 (0.036) 0.657 (0.419) 0.158 (0.064)

-0.529 (0.036) 2.284 (0.097) 19.489 (0.970)

-0.011

-0.094

-0.013

-0.085

-0.021

0.015

-0.094

-0.087

0.008 115.787

0.163 23.613

0.005 61.292

0.173 1.193

0.001 37.745

0.061 12.008

0.329 2.142

0.5 3.552

Numbers in parenthesis represent statistical significance level. MS-VECM = Markov switching vector error correction model; ECT = error correction term. 3 (1) and (2) refer to the used models. 2

International Food and Agribusiness Management Review

122

Popovic et al.

Volume 20, Issue 1, 2017

presented in Table 4, is around 0 in all four markets for the standard regimes and is significantly different than 0 in cases of extreme regimes. Regarding long-run price transmission, the parameter results show the relative differences of long-run slope coefficients between standard and extreme regimes in all cases. New Zealand, USA and Serbia have an increasing trend of elasticity in the extreme regimes, when compared to the standard regimes. For New Zealand and USA, the long-run price elasticities increase in extreme regimes compared with standard for 118 and 49%, respectively. Those markets react faster to extreme milk price changes on the world market. The long-run price elasticity in Germany decreases for 56% in the case of the extreme regime, which can be explained by the fact that milk prices on the national market are mostly above world milk price levels and they are more stable. In Serbia, the long-run price elasticity strengthens by 18% in the extreme regime, which means that the Serbian market ‘modestly’ reacts to world milk price changes, compared to the other three markets. This study also identified highly significant contemporaneous price transmission parameters in the Markov switching models for Serbia, Germany and New Zealand. Those contemporaneous price transmissions are higher in the extreme regimes than in the standard regimes. The speed of adjustment of deviations from the long-run equilibrium is statistically significant in each regime switching model, showing that the examined markets are integrated with each other and to the world dairy market, while New Zealand has the strongest integration. The larger the domestic price changes, the higher the speed of adjustment (Götz et al., 2013). The equilibrium between the national market and the world market is illustrated by the level of the error correction term (ECT). In other words, it presents the size of deviation from the long-run equilibrium between the national and the world market. This term is calculated as follows: national – β – β ln p world (3) ECTt = ln pt-1 0 1 t-1

The calculated error correction term signifies that when the size of the absolute value is larger, greater disconnection between actual prices and the equilibrium level emerges. An average ECT>0 depicts that the national price is higher than its equilibrium level; while average ECT<0 indicates that the national price is lower than its equilibrium level. Thus, when the average ECT<0 producers suffer losses due to the unfavourable price development and compared to the equilibrium state, especially in case of extreme regimes in Serbia, USA and Germany. The market stability is shown by the regime-specific standard error of the estimated model. The estimated standard errors for the extreme regime are considerably higher than standard errors for the standard regime for Serbia, Germany and New Zealand. The USA market has substantially higher standard errors in both regimes compared to standard errors of the other three markets in the standard regime. Such market price volatility increases market uncertainty along with a pessimistic aftermath on investment perspectives. The duration of the domestic market effects could confirm the previous conclusion about market stability. In the case of Serbia, Germany, and New Zealand, the average duration of the standard regime is significantly higher than the average duration of the extreme regime. It means that after the extreme regime is over, the standard regime remains for a long period of time. In other words, the standard regime pushed the market closer to its equilibrium and decreased the level of market instability. The applied MS-VECM model is used to estimate farm gate milk prices in Serbia, if one assumes a scenario without the milk crisis (Figure 4). A similar approach for the wheat sector in Serbia was used by Djuric et al. (2012). The first scenario is a dynamic forecasting from the presented model, which uses only forecasted values of the lagged dependent variables. The second scenario uses static forecasting, taking in account actual values of the lagged dependent variables. The first scenario offers an average forecast estimation according to the domestic market situation and expectations, while the second scenario involves more influence of international dairy price changes. Based on such an estimate, it is possible to give answer to the question how much did dairy farmers lose during the aflatoxin crisis? Both scenarios respond with significantly higher International Food and Agribusiness Management Review

123

Price (EUR/100 kg ECM)

Popovic et al.

45 40 35 30 25 20 15 10 5 0 Jan-07

Volume 20, Issue 1, 2017 World market price (IFCN)

Jan-08

Jan-09

Jan-10

Farm gate price

Jan-11

Scenario 1

Jan-12

Jan-13

Scenario 2

Jan-14

Serbia

Figure 4. Milk price estimate for Serbia with two scenarios without Aflatoxin M1 crisis (adapted from IFCN, 2015; OANDA, 2015; RZS, 2014a). ECM = energy-corrected milk. prices than actual during crisis period. Multiplying the average monthly milk production amount, during the 23 month period, by the milk price differences (forecasted minus actual), infers that Serbian dairy farmers suffered a loss of 96.2 million EUR in Scenario 1 and 74.7 million EUR in Scenario 2. Serbian farmers, who produce and deliver at least 3,000 litres of milk per quartile, were able to receive milk premiums during the crisis. This coupling measure is one of most important single measures for dairy farmers in agricultural policy. Although milk premiums were questioned before the crisis regarding their level and how they are paid, they remained and helped farmers that delivered milk to dairy plants to partially cover their losses. In total, during the crisis farmers received 77.45 million EUR of milk premiums. It can be inferred that the loss incurred during the crisis is significantly covered by this measure. But, the distribution of premiums was highly unequal. Only one smaller group of dairy farms, more precisely, medium sized and large farms, received premiums during 2013 and 2014. Their shares in total number of dairy farms during years of crises were 10.5 and 11.8% respectively. The vast number of small farms with only 3 or 4 cows, depending of production region, didnâ&#x20AC;&#x2122;t receive such support from the government. Those farms produce half of all milk in Serbia. Besides that, small farms receive significantly lower milk prices from dairy companies and have difficulty improving their milk quality. Therefore, the crisis hit the small farms hardest. The loss suffered during the crisis significantly slowed down investment in dairy farming, especially on the numerous small farms. At same time, production costs during the crisis increased for all farms because of the use of mycotoxin absorbers and additional milk testing. On subsequent link of dairy supply chain, processors coped with crisis on various ways. Some, mostly bigger dairy processors diminishing margins and organise periodical sales with retailers, while others built their own retail network of small dairy shops, focusing more on alternative market channels or niche markets, etc. No doubt, it can be inferred that part of the cost of the crisis were carried by processors, but exact details of this were not gathered in this study.

6. Conclusions This paper examined a problem with Aflatoxin M1 in the Serbian dairy industry and its impact on the domestic market. The empirical results indicate the occurrence of two different states of milk market integration i.e. price transmission regimes, for: Serbia, Germany, New Zealand and USA. Serbian milk market is integrated in the world milk market, although Serbian farmers do not benefit fully from high milk price changes, but it did not fully suffer from low world milk prices either. Compared with the other three markets in the longrun, the Serbian market reacted only partly to world milk price changes. Serbia has market stability in the

International Food and Agribusiness Management Review

124

Popovic et al.

Volume 20, Issue 1, 2017

standard regime, but in the extreme regime the market is destabilized, much like in Germany, which can be explained by the geographical proximity of these markets. The model didn’t confirm an extreme regime in Serbian dairy sector during the Aflatoxin crisis, as was confirmed in several recent studies for other agricultural products during periods of market disturbances (Gotz et al., 2013; Djuric et al., 2012). Specific characteristic of cow milk production, with its long biological lag, is one reason why milk processing companies didn’t react to protect short-term profit. The dairy sector in Serbia is an exceptionally good example of vertical coordination during the last two decades. Big dairy companies first start to work with dairy farmers, followed recently by middle sized and small milk processors. Vertical coordination among farmers and milk processors helped both sides to alleviate challenges brought by crisis. Simulated scenarios confirm occurrence of loss during the aflatoxin outbreak in the Serbian milk market. The estimated range of the loss was between 74.7 and 96.2 million EUR in the period from January 2013 to December 2014. It infers that the aflatoxin crises temporally reduced the degree of Serbian market integration with world milk markets, causing losses for the national dairy supply chain. The key lessons learned from the milk aflatoxin crisis can be presented in two segments. The first group covers effects on the food supply chain, and the second to the government response to the food scare crisis. The burden of the crisis was not distributed proportionally among farmers. Middle and large dairy farms (22%) that produce half of milk on national market, were protected from two sides. Dairy processors typically pay significantly higher milk prices to those farms to secure quantity and quality inputs. Additionally, dairy subsidies, retained during the crisis, in total were almost enough to compensate the dairy farm’s losses, but they were allocated only to middle and big size farms. That helped them to overcompensate for its forecasted loss during the crisis. On the other side, the numerous small dairy farms, representing second half of national market, bore the biggest burden of negative crisis effects. Small dairy farms were not able to compensate for their loss with milk premiums, and received significantly lower milk prices. Their investment abilities were significantly reduced. In the period of preparation for EU accession, such investment on small dairy farms is crucial for their future. Experience from the milk contamination crisis shows once more the importance of vertical cooperation between processing companies and family farms in developing countries, especially those without cooperatives. Dairy processing companies were under pressure and lost profit during crisis. Without market power, transferred to highly concentrated retailers about decade ago, dairy companies were challenged to manage the situation alone and to protect own long-term interest. Beside loss of some traditional export markets, dairy companies were looking for opportunities in the new markets and succeeded to lower the supply pressure on the domestic market. The aflatoxin milk crisis in Serbia reveals typical mistakes in crisis management. The lack of credible information, daily accusations, and dodging responsibility between government and its opposition, together with silence from the dairy companies and scientists were the causes of expanding negative effects. The food crisis proved to be a very ‘seismic active’ area for government. The two year crisis was managed by three ministers of Agriculture. ‘Easy’ policy measures like a tenfold increase in allowed level of aflatoxin in milk proved to be highly negative and further decreased demand. Besides that, the crisis brought unexpected cost for the government, because of the retained milk premiums during the crisis. Negative experience forced government to negotiate with food supply representatives for a Protocol of cooperation in communication during food scare crises, immediately after the crisis. The contribution of research in this paper to literature on food scare crisis is composed from linking the milk aflatoxin M1 crisis and long-run milk price transmission on the Serbian milk market, and examined approach how to measure the crisis’ effect on the most vulnerable link in the milk supply chain, the farmers. Also, the distribution of the crisis burden among farm types was examined. Additionally, adjusting government International Food and Agribusiness Management Review

125

Popovic et al.

Volume 20, Issue 1, 2017

policy during the crisis as well as adjusting of other links in the milk supply chain were analysed. An open question for future research is how to design procedures for measurement of food scare crisis effects on each food chain link for effective policy measures.

Acknowledgments This study was supported by the Ministry of Education, Science and Technological Development of the Republic of Serbia, Projects No. III 46009 and III 46006.

References Anderson, D.P. and J.D. Anderson. 2009. Food price inflation in livestock product markets: from live animal to retail prices. International agricultural trade research consortium analytic symposium ‘confronting food price inflation: implications for agricultural trade and policies’. Available at: http://tinyurl.com/ hgdrqfl. Ardeni, P.G. 1989. Does the law of one price really hold for commodity prices? American Journal of Agricultural Economics 71: 661-669. Balcombe, K., A. Bailey and J. Brooks. 2007. Threshold effects in price transmission: the case of Brazilian wheat, maize and soya prices. American Journal of Agricultural Economics 84: 292-307. Banati, D. 2011. Consumer response to food scandals and scares. Trends in food science and technology 22: 56-60. Barrett, C.B. and J.R. Li. 2002. Distinguishing between equilibrium and integration in spatial price analysis. American Journal of Agricultural Economics 89: 308-323. Bergeaud-Blackler, F. and M.P. Ferretti. 2006. More politics, stronger consumers? A new division of responsibility for food in the European Union. Appetite 47: 134-142. Brummer, B., S. von Cramon-Taubadel and S. Zorya. 2009. The impact of market and policy instability on price transmission between wheat and flour in Ukraine. European Review of Agricultural Economics 36: 203-230. Djuric, I., L. Gotz and T. Glauben. 2012. Global commodity price peaks and governmental interventions: the case of the wheat-to-bread supply chain in Serbia – did consumers really benefit? 52nd Annual conference of the German society of economics and social sciences in agriculture. Available at: http://tinyurl.com/jhzfozo. Fackler, P.L. and B.K. Goodwin. 2001. Spatial price analysis. In: Handbook of agriculture economics, edited by B. Gardner and G. Rauser. Elsevier Science, Amsterdam, the Netherlands, pp. 971-1024. Ferris, J. 1997. Agricultural prices and commodity market analysis. Michigan State University Press, East Lansing, MI, USA. Food and Agriculture Organization of the United Nations (FAOSTAT). 2015. Database. Available at: http:// www.fao.org/faostat/en. Goodwin, B.K. and N.E. Piggott. 2001. Spatial market integration in the presence of threshold effects. American Journal of Agricultural Economics 83: 302-317. Gotz, L., T. Glauben and B. Brummer. 2013. Wheat export restrictions and domestic market effects in Russia and Ukraine during the food crisis. Food Policy 38: 214-226. Greene, W.H. 2003. Econometric analysis, 5th edition. Prentice Hall, New Jersey, NJ, USA. Hamilton, J.D. 1989. A new approach to the econometric analysis of nonstationary time series and business cycle. Econometrica 57: 357-384. Hassouneh, I., A. Radwan, T. Serra and J.M. Gil. 2012. Food scare crises and developing countries: the impact of avian influenza on vertical price transmission in the Egyptian poultry sector. Food policy 37: 264-274. Hassouneh, I., T. Serra and J.M. Gil. 2010. Price transmission in the Spanish bovine sector: the BSE effect. Agricultural Economics 41: 33-42. Hemme, T. 2008. Milk equivalent concept. IFCN Dairy report 2008, International farm comparison network. IFCN Dairy research center, Kiel, Germany, p. 68. International Food and Agribusiness Management Review

126

Popovic et al.

Volume 20, Issue 1, 2017

Hemme, T. 2014. IFCN Dairy report 2014. International farm comparison network. IFCN Dairy research center, Kiel, Germany. Hussaini, A.M. 2013. Mycotoxin and food safety in developing countries. InTech, Rijeka, Croatia. International Farm Comparison Network – Dairy Research Center (IFCN). 2015. Monthly world milk prices, January 2015. Available at: www.ifcndairy.org. Kos, J., J. Levic, O. Djuragic, B. Kokic and I. Miladinovic. 2014. Occurrence and estimation of aflatoxin M1 exposure in milk in Serbia. Food Control 38: 41-46. Kos, J., J. Mastilovic, J.E. Hajnal and B. Saric. 2013. Natural occurrence of aflatoxins in maize harvested in Serbia during 2009-2012. Food Control 34: 31-34. Krolzig, H.M. 1997. Markov-switching vector autoregression – modelling, statistical inference, and application to business cycle analysis. Springer, Berlin, Germany. Lloyd, T., S. McCorriston, C. Morgan and A. Rayner. 2001. The impact of food scares on price adjustment in the UK beef market. Agricultural Economics 25: 347-357. Mitchell, L. 2003. Economic theory and conceptual relationships between food safety and international trade. In: International trade and food safety economic theory and case studies, edited by J.C. Buzby, ERS USDA. Available at: http://tinyurl.com/j3eqerw. Motarjemi, Y. and H. Lelieveld. 2014. Food safety management, a practical guide for the food industry. Elsevier, Oxford, UK. OANDA. 2015. Currency converter, historical exchange rates. Available at: http://tinyurl.com/hjnuq6a. Phoong, S.W., M.T. Ismail and S.K. Sek. 2014. Linear vector error correction model versus Markov switching vector error correction model to investigate stock market behaviour. Asian Academy of Management Journal of Accounting and Management 10: 133-149. Popovic, R. and B. Radovanov. 2010. Price transmission in Serbian milk commodity chain. Economics of agriculture 57: 543-554. Popovic, R., B. Radovanov and M. Jeremic. 2013. Volatile world milk prices and its affect to national market – case of Serbian milk market. In: Thematic proceedings 135th EAAE seminar, edited by D. Tomić, M. Sevarlic, K. Lovre and S. Zekic. Available at: http://tinyurl.com/jnveo3l. Skrbic, B., J. Zivancev, I. Antic and M. Godula. 2014. Levels of aflatoxin M1 in different types of milk collected in Serbia: assessment of human and animal exposure. Food Control 40: 113-119. Statistic office of Republic of Serbia (RZS). 2013. Census of agriculture 2012 – agriculture in the Republic of Serbia, Available at: http://tinyurl.com/zcf226g. Statistic Office of Republic of Serbia (RZS). 2014a. Economic accounts for agriculture, in the Republic of Serbia 2007-2013. Available at: http://webrzs.stat.gov.rs Statistic Office of Republic of Serbia (RZS). 2014b. Household budget survey 2013. Available at: http:// webrzs.stat.gov.rs. Terasvirta, T. 1994. Specification, estimation and evaluation of smooth transition autoregressive model. Journal of the American Statistical Association 89: 208-218.

International Food and Agribusiness Management Review

127

OPEN ACCESS International Food and Agribusiness Management Review Volume 20 Issue 1, 2017; DOI: 10.22434/IFAMR2016.0090 Received: 3 May 2016 / Accepted: 22 September 2016

Food safety and food imports in Europe: the risk of aflatoxins in pistachios RESEARCH ARTICLE Bo Xiong Senior Research Analyst, Agricultural Issues Center, University of California, One Shields Avenue, UC-Davis, CA 95616, USA

Abstract The United States has surpassed Iran as the largest pistachio exporter to the European Union. Both lower prices and a less frequency of aflatoxin contamination have contributed to the success of the US pistachio industry. Using EU monthly imports and food safety alerts data, we estimate EU demand for US and Iranian pistachios. We find that EU demand for US pistachios is price-inelastic but the demand for Iranian pistachios is price-elastic. We also find that the income effect is positive for US nuts but negative for Iranian nuts. Most importantly, we find that EU imports of US pistachios decrease with aflatoxin alerts traced back to the US but increase with contamination incidents originated from Iran. Keywords: pistachio, tree nut, aflatoxin, food safety, trade JEL code: Q17, Q18 Corresponding author: boxiong@ucdavis.edu; bonapartexiongbo@gmail.com

ÂŠ 2017 Xiong

129

Xiong

Volume 20, Issue 1, 2017

1. Introduction The world market for pistachios The average world consumption of pistachios is 475,000 metric tons per year over the period 2009-2012 (Pistachio PSD USDA-FAS, 2016). The United States and Iran are the top two producers in the world. Specifically, the United States supplies 180,000 tons (or 38%) annually and Iran produces 160,000 tons (or 34%) per year.1 Other major producing nations are Turkey and Syria, jointly supplying 130,000 tons (or 28%). Because of the high concentration of production, international trade is important to the pistachio market. Nearly 200,000 to 300,000 tons of pistachios are traded across national borders each year (excluding intraEU trade). The United States and Iran are the dominant exporting countries, accounting for 50 and 45% of world exports, respectively. Among various destination markets, the EU is the world’s largest importing region at 30% of the world’s total imports. Changes in the EU market The United States has surpassed Iran as the largest pistachio supplier to the EU. As shown in Figure 1, US monthly pistachio exports to the EU increased from 1000 tons in 1999 to nearly 3,500 tons in 2012. During the same period, Iranian exports to the EU declined from 4,000 tons to less than 1,500 tons per month. What explains the rise of US and the decline of Iran in the European pistachio market? The change in relative prices is one reason. As shown in Figure 2, US pistachios have become more competitive than Iranian pistachios in recent years. The comparative advantage of US pistachios is primarily driven by higher yields in California as a result of improved management practices. In particular, Geisseler and Horwath (2016) report that Californian pistachio yield has steadily risen to 3,541.9 kg/ha as of 2012, while Iranian pistachio yield remains at 1,814.7 kg/ha according to FAOSTAT (2016). 1 The

production quantities are averaged over four consecutive years, 2009-2012, because pistachio trees feature alternate bearing.

8 Iran

×1000 metric tons

6 5 4 3 2

Jul-12

Jan-13

Jul-11

Jan-12

Jul-10

Jan-11

Jul-09

Jan-10

Jul-08

Jan-09

Jul-07

Jan-08

Jul-06

Jan-07

Jan-06

Jul-05

Jan-05

Jul-04

Jan-04

Jul-03

Jan-03

Jul-02

Jan-02

Jul-01

Jan-01

Jul-00

Jan-00

Jan-99

Jul-99

Figure 1. EU monthly imports of pistachio from Iran and United States, 1000 metric tons (adapted from EUROSTAT, http://ec.europa.eu/eurostat). International Food and Agribusiness Management Review

130

Xiong

Volume 20, Issue 1, 2017 10 Iran

Price (€/kg)

8 7 6 5 4 3

Jul-12

Jan-13

Jul-11

Jan-12

Jul-10

Jan-11

Jul-09

Jan-10

Jul-08

Jan-09

Jul-07

Jan-08

Jul-06

Jan-07

Jan-06

Jul-05

Jan-05

Jul-04

Jan-04

Jul-03

Jan-03

Jul-02

Jan-02

Jul-01

Jan-01

Jul-00

Jan-00

Jul-99

Jan-99

Figure 2. EU monthly import prices of Iranian and US pistachios in €/kg. Import prices are the cost, insurance and freight prices measured as the unit-values of imported pistachios at EU customs (adapted from EUROSTAT, http://ec.europa.eu/eurostat). Nevertheless, the US price advantage is no more than €1 per kilogram based on Figure 2.2 Therefore, nonprice factors such as food safety issues presumably play an important role in driving the consumption trends in Europe. Aflatoxin control in the European Union and the United States Aflatoxins are naturally occurring substances produced by certain fungi growing on nut products, maize, and other agricultural commodities. The fungi are more likely to develop under hot and humid conditions. Acute intoxication through aflatoxin-contaminated food can lead to liver damages or developmental delays, and feedstuff spoiled with aflatoxin causes weight loss in animals. Aflatoxin control in pistachios is an important food safety measure in the EU. In 1997, the European Commission (EC) temporally suspended imports of Iranian pistachios because of severe aflatoxin contaminations (EC, 1997a). Although the ban was repealed later in the year, restrictive measures were imposed on pistachio imports from Iran, including mandatory sampling and testing procedures at customs (EC, 1997b). In April 2002, the EC set the union-wide maximum residue limits (MRLs) for the aflatoxin content in food and feed (EC, 2001, 2002). The EU MRL is 10 μg/kg for pistachios for direct human consumption and 15 μg/kg for pistachios subject to further processing. In December 2006, the EC modified some MRLs but the MRLs for aflatoxins in nut products remained (EC, 2006). In November 2009, due to the high risk of aflatoxin contamination, the EC imposed more rules governing the imports of specified products from specified countries, including pistachio and derived products from Iran and Turkey (EC, 2009). Figure 3 shows that aflatoxin contamination in pistachios is more severe in Iran than in the US From 1999 to 2012, there were a total of 98 EC alerts concerning aflatoxin-spoiled pistachios traced back to Iran. In comparison, only 21 alerts involved US pistachios during the same period. Therefore, aflatoxin control measures in the EU can be a key factor that changes the landscape of international pistachio markets.

2 In Supplementary Table S1 we show that the price of US pistachios was higher than that of Iranian pistachios over the period 1999-2007. However,

the price of US pistachios was lower than that of Iranian pistachios by 1 euro/kg over the period 2008-2012. Both test results are statistically significant.

International Food and Agribusiness Management Review

131

Xiong

Volume 20, Issue 1, 2017 100 90 80

Iran US

Number of alerts

70 60 50 40 30 20 10 Jan-99 May-99 Sep-99 Jan-00 May-00 Sep-00 Jan-01 May-01 Sep-01 Jan-02 May-02 Sep-02 Jan-03 May-03 Sep-03 Jan-04 May-04 Sep-04 Jan-05 May-05 Sep-05 Jan-06 May-06 Sep-06 Jan-07 May-07 Sep-07 Jan-08 May-08 Sep-08 Jan-09 May-09 Sep-09 Jan-10 May-10 Sep-10 Jan-11 May-11 Sep-11 Jan-12 May-12 Sep-12 Jan-13

Figure 3. EU accumulative alerts of aflatoxin-contaminated pistachios imported from Iran and United States (adapted from RASFF, 2016). Aflatoxin policy in the United States has also been enhanced over the past decade. The US had been using 20 Îźg/kg as the MRL for aflatoxin content in all foods (except for milk) since 1969 (FAO, 2004). In August 2005, a federal marketing order (FMO) initiated by the US pistachio industry took effect.3 The FMO requires mandatory testing for aflatoxin content for all pistachios marketed in the United States, Canada, and Mexico.4 The effective MRL for aflatoxins in the United States is 15 Îźg/kg, which is as stringent as the EU standard. Gray et al. (2005) project that the benefit of the FMO, through enhancing consumer demand, should outweigh the cost of the program. Because the aflatoxin standards implied by the FMO are less stringent than the EU standards, we stipulate that the FMO has negligible impacts on nut importers in Europe. Summary of key findings We contribute to the understanding of food import and food safety by offering a case study of the EU pistachio market. The market is unique in that the supply is dominated by two countries: United States, a developed nation, and Iran, a developing nation. Therefore, our case study is an ideal experiment to investigate the distributional effects of food safety policies on agricultural exporters from countries of different development stages. Using EU monthly import and food safety alert data from 1999 to 2012, we estimate EU demand for US and Iranian pistachios. We find that EU demand for US pistachios is price-inelastic but the demand for Iranian pistachios is price-elastic. We also find that the income effect is positive for US nuts but negative for Iranian nuts. Most importantly, we find that EU imports of US pistachios decrease with aflatoxin alerts traced back to US pistachios but increase with contamination incidents originated from Iran. In particular, we estimate that ten more food safety alerts involving US pistachios would reduce EU monthly imports of US nuts by 167 tons, while ten more alerts concerning Iranian pistachios would raise EU imports of US nuts by 200 tons a month. 3 http://tinyurl.com/z9nmhsz.

4 Pistachio production in North America is concentrated in San Joaquin Valley and surrounding areas in California. Over the past decade, the region

has invested heavily in tree nuts such as almond, pistachios, and walnut, as well as fruits such as grapes and cherries. Traditionally, the agricultural sector within the region was dominated by dairy farms and cattle ranches.

International Food and Agribusiness Management Review

132

Xiong

Volume 20, Issue 1, 2017

The rest of the article is organized as follows. In Section 2 we introduce the empirical specification to characterize the EU imports of pistachios. We present the data and their sources in Section 3 and discuss the econometric results in Section 4. In Section 5 we offer concluding remarks and suggest directions for future research.

2. EU demand for pistachios Literature on food safety and food import Previous research on the impact of food safety policy, or non-tariff measures in general, often uses the gravity equation approach.5 Otsuki et al. (2001) projected that the harmonization of aflatoxin policies within the EU would reduce Africa’s exports of cereals, dried fruits and nuts by 64%. Disdier et al. (2008) find that the sanitary and phyto-sanitary measures and technical barriers to trade significantly constrain developing countries’ agricultural exports to the OECD markets. Munasib and Roy (2011) report that maize exporters from poor countries are negatively affected by aflatoxin control in the developed world. Xiong and Beghin (2012) find that EU aflatoxin policies have negligible impacts on groundnut exports from Africa to Europe. Winchester et al. (2012) report that cross-country differences in pesticide regulations reduces EU trade in plant products with major partners, with the magnitude of the trade distortion varying substantially across sectors. Despite its empirical success, the gravity equation approach does not attend to price effects, which are important to the pistachio market because other tree nuts can be close substitutes or complements. To characterize substitution patterns across a variety of tree nuts, we use a partial demand system to model EU imports of pistachios. In a related study, Zheng et al. (2012) empirically estimate the export of US pistachios, with food safety attributes measured by incidents from the Google news timeline. The econometric specification The EU pistachio imports respond to prices of pistachios and other types of tree nuts, the EU income level, and the EU-wide food safety alerts. Following LaFrance (1990), we specify EU import demand functions for pistachios as: qtUS = α0 + α1It + α2[ptUS (1+tartUS)] + α3[ptIran (1+tartIran)] + α4 pta + α5 pth + α6 fatIran – α7 fatUS + µt (1) qtIran = β0 + β1It + β2[ptIran (1+tartIran)] + β3[ptUS (1+tartUS)] + β4 pta + β5 pth + β6 fatIran – β7 fatUS + vt (2) where qtUS is the quantity of pistachios EU imports from the US, It is EU quarterly GDP smoothed over months, ptUS and ptIran are the EU import prices of US and Iranian pistachios, tartUS and tartIran are the ad valorem tariff rates EU imposes on pistachios of two origins, and pta and pth are EU import prices of US almonds and Turkish hazelnuts, respectively.6 By including the prices of the two alternative tree nuts, we relate EU imports of pistachios to its imports of other types of nut products. Because the US and Turkey are the dominant suppliers of almonds and hazelnuts, the two price series represent world prices.7 In Equation 1, the variable fatIran denotes the frequency of aflatoxin alerts involving pistachios imported from Iran. Specifically:

5 See Anderson and Van Wincoop (2003) for a derivation of the gravity equation. Alternatively, the price-wedge method is used to analyze non-tariff

barriers. For example, Nimenya et al. (2012) assess the impact of EU non-tariff measures on imports of horticultural and fish products from Africa. Bureau and Beghin (2001) provide a review of different methods in the analysis of non-tariff measures. 6 Note that we omit EU domestic price of pistachios because EU production accounts for 2% of world pistachio production (Pistachio PSD USDAFAS, 2016). 7 We omit the prices of other tree nuts (e.g. pecans, walnuts, Brazil nuts) because EU import of these nut products is small in quantity. We also omit EU import duties on US almonds and Turkish hazelnuts because both duty rates remain constant over the past decade.

International Food and Agribusiness Management Review

133

Xiong

Volume 20, Issue 1, 2017 t–1

fatIran =

Iran ∑ alertsm

m=1 t–1

Iran ∑ qm

(3)

m=1

Intuitively, the variable uses historical information on food safety and import to measure the extent to which a ton of Iranian pistachios is at risk of aflatoxin contamination. This novel measurement improves upon the simple count of Rapid Alert System for Food and Feed (RASFF) alerts in Jaud et al. (2013) by addressing the endogeneity problem of the count variable.8 That is, with food safety risk held constant, a product traded in higher volumes triggers more alerts because the base for sampling and testing is larger. It is also worth noting that we implicitly assume that the consignments affected by alerts are comparable in volumes. Ideally, the number of alerts should be further translated into the quantity of contaminated pistachios. Unfortunately, the RASFF database does not provide information on the volumes of the at-risk consignments. Similarly, we measure the frequency of aflatoxin contamination in US pistachios by the variable: t–1

fatUS

∑ alertsUS n

n=1 t–1

∑ qUS n

(4)

n=1

We focus on aflatoxin incidents because they are the primary food safety concern in pistachio nuts. The empirical specification Equation (1)-(2) has two major advantages. First, the measurement of the severity of food safety incidents is normalized by the size of import. This novel approach allows us to disentangle the food safety effect from the market size effect. Second, the specification explicitly controls for the substitution patterns between pistachios and alternative tree nuts. Therefore, the forthcoming results are informative to stakeholders in the business of tree nuts in general.

3. The data The custom data Using the EUROSTAT database (http://ec.europa.eu/eurostat), we retrieve monthly imports of pistachios, from Iran and the United States, respectively, for the 27 EU member states from January 1999 to December 2012.9 We start the period of examination from January 1999 for two reasons. First, the EU banned Iranian pistachios in 1997 and applied highly restrictive measures against Iranian nuts in 1998. Second, the import data are available in both value (euro) and quantity (kilogram) since January 1999. We compute the import prices of Iranian and US pistachios by taking the ratio of the import values over the import quantities. The unit-value measurement of prices is likely subject to measurement errors. Specifically, any potential error in the quantity series would introduce a negative correlation between the price and quantity measurements, resulting in biased estimates. After reviewing the data, however, we stipulate that the magnitude of the potential bias should be limited for two reasons. First, random errors due to rounding are unlikely because the value and quantity series are recorded down to the last digit of euro and kilogram. Second, the data aggregation from the port level to the EU level mitigates idiosyncratic measurement errors at individual customs. In a similar way, we compute EU monthly import prices of US almonds and Turkish hazelnuts.

8 Similar

to Jaud et al. (2013) and Piggott and Marsh (2004) use the count of articles published in major newspapers to measure the safety of meat products in the United States. 9 Since January 2012 the Harmonized System further classifies pistachios into in-shell pistachios (HS 080251) and shelled pistachios (HS 080252). Over 90% of EU imports of pistachios are in-shell nuts. We aggregate the two categories into in-shell equivalence.

International Food and Agribusiness Management Review

134

Xiong

Volume 20, Issue 1, 2017

The rapid alert system for food and feed data The RASFF of the EU informs member states of the detected risks related to food and feed products.10 There are two types of RASFF alerts: market notifications and border rejections. A market notification is triggered when the identified risk requires rapid actions from all member states (e.g. withdrawal of the product). A border rejection denies the entry of the consignment for safety, labeling, or other specification issues. In the empirical analysis we focus on market notifications because they are directly relevant to the entire EU market. We collect all RASFF market notifications concerning aflatoxin-contaminated pistachios or derived products, imported from Iran and the US For example, on January 6th 2011, the United Kingdom identified that some unsalted pistachio nuts imported from the United States contained aflatoxin B1 above the EU MRL. Consequently, the RASFF issued a market notification requiring all member states to withdraw the products from the marketplace. Admittedly, the RASFF market notification data do not fully reflect the severity of food, safety problems for two reasons. First, RASFF does not disclose the volume or the value of an affected shipment. Second, RASFF does not provide information on the extent to which the EU standards are violated.11 Nevertheless, RASFF remains a systemic and up-to-date databank for food safety issues. The tariff and exchange rate data We use the EU effectively applied tariff rates from the TRAINS database of the World Bank (http://tinyurl. com/zsfnckj). The ad valorem rates faced by Iran and the US are available for 2008 and 2009. We complement the TRAINS tariff data with current duty rates listed in the EC TARIC database.12 The EU third-country rate for pistachios is 1.6%. We extract monthly real exchange rates from the Economic Research Service of the US Department of Agriculture.13 We express exchange rates in terms of per euro. In Supplementary Table S2, we present summary statistics of all variables used in the estimation.

4. Results and discussions Model validation checks ■■ Unit root tests Since we use monthly data spanning from 1999 to 2012, we first check the stationarity of EU import quantities of pistachios. The Dickey-Fuller unit root test results suggest that neither the import quantity of Iranian pistachios nor that of US nuts contains unit roots.14 Therefore, the system Equation (1)-(2) can be estimated in its original form as opposed to in its differenced specification. ■■ Endogeneity of import prices The endogeneity of pistachio prices is another caveat in the empirical estimation. In recent years, the EU imports from Iran accounts for nearly 15% of all Iranian pistachio exports and EU imports from the US amount to 27% of all US pistachio exports (Pistachio PSD USDA-FAS, 2016). Therefore, any unobservable determinants of EU import demand are likely to affect the world prices of pistachios as well. To check the

10 The

RASFF data portal can be accessed at http://tinyurl.com/zaer5l8. an illustration, a parcel of pistachios tested 16 μg/kg enters into the RASFF system in the same way as a parcel of pistachios tested 50 μg/kg. 12 The exchange rate data is available from http://tinyurl.com/jjqgzod. 13 The data is available at http://tinyurl.com/jaey6rq. Because the real exchange rate for the Iranian rial is unavailable on a monthly basis, we use the annual estimates as an approximation. 14 The associated test results are available in Supplementary Table S3. 11 As

International Food and Agribusiness Management Review

135

Xiong

Volume 20, Issue 1, 2017

exogeneity of EU import prices of pistachios, we conduct the Hausman-Durbin-Wu tests with monthly real exchange rates (i.e. dollar per euro and rial per euro) as the instrumental variables.15 The Hausman-Durbin-Wu test results suggest that EU import price of US pistachios can be taken as exogenous but EU import price of Iranian pistachios is endogenously determined. To address the endogenous price of Iranian pistachios, we combine the system Equation (1)-(2) with a third equation that explains EU import price of Iranian pistachios. Specifically, we use all exogenous variables including the two exchange rates as the potential determinants of the import price of Iranian nuts. ■■ The properties of demand To ensure that our empirical specification satisfies Marshallian symmetry, we impose the parameter restriction α3=β3. Note that we use Marshallian symmetry to approximate Hicksian symmetry because the expenditure on pistachios accounts for a tiny share in the overall budget (LaFrance, 1990). To retain the homogeneity of degree zero in all nominal terms, we deflate all monetary measurements by EU monthly Consumer Price Indices (CPI).16 The benchmark results We use the three-stage least square procedure to estimate the system Equation (1)-(2) in which the price of Iranian pistachios is endogenously determined. Table 1 displays the estimation results. As shown in Table 1, the EU’s demand for US pistachios decreases with the price of US pistachios and increases with the price of Iranian pistachios. The substitutability between US and Iranian pistachios is also reported by Zheng et al. (2012). We also find that the demand for U.S pistachios rises with the income level in Europe. Furthermore, we find that US pistachios are complements to US almonds but substitutes for Turkish hazelnuts. We later discuss the relations among various tree nuts in more detail. Most importantly, 15 The 16 For

results of the Hausman-Durbin-Wu tests are available in Supplementary Table S4. example, if all prices double, the CPI doubles as well. Therefore, all real prices are unchanged, so is the demand.

Table 1. Regression results for EU imports of pistachios, 1999-2012.1,2

Import price of US pistachio Import price of Iranian pistachio Real exchange rate (dollar/euro) Real exchange rate (1000 rial/euro) EU income Import price of US almond Import price of Turkish hazelnut Frequency of aflatoxin incidents, traced to US Frequency of aflatoxin incidents, traced to Iran Constant R2

Import quantity of US pistachio

Import quantity of Iranian pistachio

Import price of Iranian pistachio

-34.08*** (11.70) 32.40*** (10.17) n.a. n.a. 0.535*** (0.075) -45.83*** (11.23) 36.37*** (10.48) -7.768*** (2.674)

32.40*** (10.17) -110.6*** (11.89) n.a. n.a. -0.189*** (0.062) 49.20*** (9.778) -29.93*** (9.632) -3.086 (2.360)

0.491*** (0.051) n.a. -0.016*** (0.003) -0.002*** (0.000) -0.000 (0.000) -0.050 (0.070) 0.288*** (0.058) 0.053*** (0.0168)

2.320 (2.408)

-0.056*** (0.018)

11.27*** (1.628) 0.71

0.083*** (0.013) 0.81

9.308*** (2.765) -13.07*** (1.988) 0.61

The Hausman-Durbin-Wu test suggests that EU import price of Iranian pistachios is endogenous. Therefore, the price variable is instrumented by monthly exchange rates of dollar and rial relative to euros. 2 Standard errors are in parentheses; *** denotes significance levels of 1%; n.a. = not available. International Food and Agribusiness Management Review

136

Xiong

Volume 20, Issue 1, 2017

we find that EU demand for US pistachios is significantly affected by food safety hazards. In particular, aflatoxin contamination incidents traced back to US pistachios significantly reduce EU imports of US nuts, and similar incidents originated from Iran enhance EU demand for US pistachios. In an earlier study of US pistachio export, Zheng et al. (2012) also test the food safety effects but their empirical analysis does not confirm the trade-impeding effect of food safety incidents. Turning to EU imports of Iranian pistachios in Table 1, we find that the EU demand for Iranian nuts decreases with the own price and increases with the price of the US pistachios. We also find that EU consumption of Iranian pistachios declines as the income level rises. In addition, we find that Iranian pistachios substitute with US almonds but complement Turkish hazelnuts. We discuss these patterns in detail after we derive the associated cross-price elasticities. In terms of food safety effects, the estimates suggest that the EU imports of Iranian pistachios do not respond to aflatoxin incidents in a statistically significant way. Finally, we attend to the price equation in Table 1. We find that the price of Iranian pistachios is positively correlated with the price of US pistachios. As expected, a weaker rial makes the Iranian nuts more competitive. However, we also find that a weaker dollar improves the competitiveness of the Iranian nuts. This counterintuitive result might have to with the role of US dollar as international currency.17 Furthermore, we find that the price of Iranian pistachios moves together with the price of Turkish hazelnuts. Most importantly, we show that Iranian pistachios are sold as discounted prices when the nuts suffer from food safety incidents, and gain price premiums when the US nuts are exposed to aflatoxin risks. This result is consistent with the notion that the attribute of food safety is partially reflected in market prices.18 Using the estimated coefficients in Table 1, we derive the associated elasticities and report them in Table 2. We find that EU demand for US pistachios is price-inelastic, with the own price elasticity at 0.5. In contrast, EU demand for Iranian pistachios is sensitive to price variations, with the own price elasticity at 4. This high responsiveness can be explained by two plausible factors. First, EU imports of Iranian pistachios declined to record low in 2011-2012 (Figure 1). Therefore, the elasticity evaluated at the low trade volume is high by definition. Second, the sensitivity to price changes of the Iranian nuts partially reflects European consumersâ&#x20AC;&#x2122; long-held risk perception before the launch of RASFF in 1979.19 In terms of the income effects, we find from Table 2 that the EUâ&#x20AC;&#x2122;s income elasticity for US pistachios is over 4, much higher than the income elasticity for most agricultural commodities (which often ranges between zero 17 One

possible explanation is that a weak US dollar indicates a sluggish world economy, which suppresses prices of commodities. Caswell and Mojduszka (1996) for a discussion of the credence nature of food safety attributes. 19 For example, Mojtahedi et al. (1979) documented the agronomic conditions that result in the development of aflatoxins in Iranian pistachios. 18 See

Table 2. Estimated elasticities for EU imports of pistachios, 1999-2012.1,2

Import price of US pistachio Import price of Iranian pistachio EU income Import price of US almond Import price of Turkish hazelnut Frequency of aflatoxin incidents, traced to US Frequency of aflatoxin incidents, traced to Iran

Import quantity of US pistachio

Import quantity of Iranian pistachio

-0.546*** (0.188) 0.597*** (0.187) 4.378*** (0.610) -0.406*** (0.099) 0.498*** (0.144) -0.129*** (0.045) 0.588*** (0.175)

1.060*** (0.333) -4.159*** (0.447) -3.156*** (1.028) 0.889*** (0.177) -0.837*** (0.269) -0.105 (0.080) 0.299 (0.310)

To reflect recent market trends, all elasticities are evaluated at the sample means in the period of marketing year 2010/2011 and 2011/2012. 2 Standard errors are in parentheses; *** denotes significance levels of 1%. International Food and Agribusiness Management Review

137

Xiong

Volume 20, Issue 1, 2017

and one). This finding suggests that US pistachio is perceived as a luxury product in Europe. Nevertheless, the extremely large income effect might also be entangled with European consumersâ&#x20AC;&#x2122; rising awareness of the health benefits of tree nuts.20 In contrast, the income effect for Iranian pistachios is negative, indicating that European consumers shift away from the products as they become more affluent. Overall, the net income elasticity for pistachios, regardless of their origins, remains positive as expected. Turning to the cross-price effects, we find from Table 2 that pistachios from the two origins are highly substitutable. We also find that pistachios substitute with other tree nuts from distant origins but complement to other nuts from surrounding regions. For example, US pistachios are substitutes with Turkish hazelnuts but complements to US almonds. Similarly, Iranian pistachios are substitutes with US almonds but complements to Turkish hazelnuts. The complementarity is unexpected and possibly because tree nut exporters to the EU are clustered by region (e.g. Middle East cluster versus North America cluster) and source various tree nuts within each region. We attend to the food safety effects in Table 2. In particular, we focus on the impacts on US pistachios because the estimated effects on Iranian pistachios are not statistically significant. We find that EU imports of US pistachios are hindered by aflatoxin incidents involving the US nuts but stimulated by similar problems traced back to the Iranian nuts. To put the estimated elasticities in perspective, ten more food safety alerts targeting US pistachios reduce EU monthly imports of US pistachios by 167 tons, while ten more alerts with reference to Iranian pistachios promote EU imports of US pistachios by 200 tons a month. The finding lends support to the hypothesis that the management of food safety risks, among other non-price factors, has contributed to the success of US pistachios in international markets. Sensitivity analysis: short-run versus long-run implications Insofar we implicitly assume away imperfect information, rigidity of contracts, or other market constraints that slow down the adjustment of monthly decisions made by EU importers. To allow partial adjustment, we conduct an alternative analysis featuring the dynamics of trade activities by using the import volume in the previous month as an additional control variable. As a side benefit, the inclusion of the import history allows us to explore the long-run implications of food safety incidents. Specifically, we re-estimate the system Equation 1-2 with the endogenous price of Iranian pistachios and with the import history as an extra control variable. For brevity, we only report the short-run elasticities and long-run elasticities derived from the new set of estimated coefficients. In particular, Table 3 displays the short-run results and Table 4 presents the long-run counterparts. As shown in Table 3 and Table 4, the elasticities are generally larger in the long term than in the short term, because current import volume is positively correlated with past import flow. The baseline results in Table 2 are more comparable with the long-run effects in Table 4. Nevertheless, the control of import history alters the statistical significance of food safety incidents. One plausible explanation is that import history already conveys information about food safety risks to a certain degree.

20 For

instance, Kris-Etherton et al. (2008) report that regular consumption of tree nuts helps prevent coronary heart disease.

International Food and Agribusiness Management Review

138

Xiong

Volume 20, Issue 1, 2017

Table 3. Estimated elasticities for EU imports of pistachios: short-run results.1,2

Import quantity of US pistachio

Import quantity of Iranian pistachio

-0.304** (0.136) 0.267** (0.122) 1.872*** (0.572) -0.193** (0.085) 0.235* (0.120) -0.057 (0.037) 0.260* (0.147)

0.473** (0.216) -1.862*** (0.368) -1.497* (0.828) 0.479*** (0.143) -0.474** (0.202) -0.120** (0.060) 0.419* (0.235)

To reflect recent market trends, all elasticities are evaluated at the sample means in the period of marketing year 2010/2011 and 2011/2012. 2 Standard errors are in parentheses; *, **, and *** denote significance levels of 10, 5, and 1%.

Table 4. Estimated elasticities for EU imports of pistachios: long-run results.1,2

Import quantity of US pistachio

Import quantity of Iranian pistachio

-0.697** (0.320) 0.610** (0.281) 4.287*** (1.147) -0.441** (0.184) 0.538** (0.262) -0.130 (0.082) 0.596* (0.324)

1.065** (0.480) -4.190*** (0.607) -3.368* (1.798) 1.078*** (0.302) -1.066** (0.447) -0.269* (0.143) 0.942* (0.557)

To reflect recent market trends, all elasticities are evaluated at the sample means in the period of marketing year 2010/2011 and 2011/2012. 2 Standard errors are in parentheses; *, **, and *** denote significance levels of 10, 5, and 1%.

5. Conclusions The United States has surpassed Iran as the largest pistachio exporter to Europe. In this article, we provide econometric evidence that the success of the US pistachio industry is partially attributable to the improved management of food safety hazards including aflatoxin risks. By estimating EU imports of pistachios from January 1999 to December 2012, we show that more aflatoxin alerts targeting US pistachios reduce EU monthly imports of US pistachios, while similar alerts concerning Iranian pistachios incentivize Europe to import more US pistachios. Our case study highlights the distributional effects of food safety policies. In particular, food safety standards and regulations are likely to put food exporters in developing countries at a disadvantage in international markets. Capital and technical assistance to producers and processors in poor nations are necessary for them to further integrate into the global market where the demand for food safety continues to rise. Future research in tree nuts can be promising in several directions. First, a demand analysis at the retail level has the potential to help agribusiness stakeholders improve their marketing strategies. It would also be interesting to investigate how consumers respond to the health benefits associated with tree nuts. In addition, the growing market for tree nuts might put a downward pressure on the market for peanuts. A relevant analysis is much needed to better inform peanut farmers in least developed countries.

International Food and Agribusiness Management Review

139

Xiong

Volume 20, Issue 1, 2017

Supplementary material Supplementary material can be found online at https://doi.org/10.22434/IFAMR2016.0090. Table S1. The price difference of US and Iranian pistachios. Table S2. Summary statistics of variables. Table S3. The Dickey-Fuller tests for EU imports of pistachios. Table S4. The Hausman-Durbin-Wu test for the exogeneity of pistachio prices.

Acknowledgements The author thanks Daniel A. Sumner for discussions, John C. Beghin for comments, and Jonathan Barker for assistance. The author also highly appreciates the suggestions from two anonymous reviewers.

Reference Anderson, J.E. and E. van Wincoop. 2003. Gravity with gravitas: a solution to the border puzzle. American Economic Review 93: 170-192. Bureau, J.-C. and J. Beghin. 2001. Quantitative policy analysis of sanitary, phytosanitary and technical barriers to trade. Ă&#x2030;conomie Internationale 3: 107-130. Caswell, J.A. and E.M. Mojduszka. 1996. Using informational labeling to influence the market for quality in food products. American Journal of Agricultural Economics 78: 1248-1253. Disdier, A.C., L. FontagnĂŠ and M. Mimouni. 2008. The impact of regulations on agricultural trade: evidence from the SPS and TBT agreements. American Journal of Agricultural Economics 90: 336-350. European Commission (EC). 1997a. Commission Decision of 8 September 1997 on the temporary suspension of imports of pistachios and certain products derived from pistachios originating in or consigned from Iran. Official Journal of the European Union L 248: 33. European Commission (EC). 1997b. Commission Decision of 11 December 1997 repealing Commission Decision 97/613/EC and imposing special conditions on the import of pistachios and certain products derived from pistachios originating in, or consigned from Iran. Official Journal of the European Union L 343: 30-34. European Commission (EC). 2001. Commission Regulation (EC) No 466/2001 of 8 March 2001 setting maximum levels for certain contaminants in foodstuffs. Official Journal of the European Union L 77: 1-13. European Commission (EC). 2002. Commission Regulation (EC) No 472/2002 of 12 March 2002 amending Regulation (EC) No 466/2001 setting maximum levels for certain contaminants in foodstuffs. Official Journal of the European Union L 75: 18-20. European Commission (EC). 2006. Commission Regulation (EC) No 1881/2006 of 19 December 2006 setting maximum levels for certain contaminants in foodstuffs. Official Journal of the European Union L 364: 5-24. European Commission (EC). 2009. Commission Regulation (EC) No 1152/2009 of 27 November 2009 imposing special conditions governing the import of certain foodstuffs from certain third countries due to contamination risk by aflatoxins and repealing Decision 2006/504/EC. Official Journal of the European Union L 313: 40-49. FAOSTAT. 2016. Crop database of Food and Agriculture Organization of the United Nations. Available at: http://tinyurl.com/hs6ct65. Food and Agriculture Organization of the United Nations (FAO). 2004. Worldwide regulations for mycotoxins in food and feed in 2003. Available at: http://tinyurl.com/z7xpvmx. Geisseler, D. and W. Horwath. 2016. Pistachio productions in California. California Department of Food and Agriculture. Available at: http://tinyurl.com/hna52qk.

International Food and Agribusiness Management Review

140

Xiong

Volume 20, Issue 1, 2017

Gray, R., D.A. Sumner, J.M. Alston, H. Brunke and A.K. Acquaye. 2005. Economic consequences of mandated grading and food safety assurance: ex ante analysis of the federal marketing order for California pistachios. Giannini Foundation of Agricultural Economics, University of California. Available at: http://tinyurl.com/hqy7k5r. Jaud, M., O. Cadot and A. Suwa-Eisenmann. 2013. Do food scares explain supplier concentration? An analysis of EU agri-food imports. European Review of Agricultural Economics 40: 873-890. Kris-Etherton, P.M., F.B. Hu, E. Ros and J. Sabaté. 2008. The role of tree nuts and peanuts in the prevention of coronary heart disease: multiple potential mechanisms. Journal of Nutrition 138: 1746S-1751S. LaFrance, J.T. 1990. Incomplete demand systems and semilogarithmic demand models. Australian Journal of Agricultural Economics 34: 118-131. Mojtahedi, H., C.J. Rabie, A. Lübben, M. Steyn and D. Danesh. 1979. Toxic aspergilli from pistachio nuts. Mycopathologia 67: 123-127. Munasib, A. and D. Roy. 2011. Sanitary and phytosanitary standards as bridge to cross (No. 1140). International Food Policy Research Institute (IFPRI). Available at: http://tinyurl.com/jzgqgjl. Nimenya, N., P.F. Ndimira and B.H. De Frahan. 2012. Tariff equivalents of nontariff measures: the case of European horticultural and fish imports from African countries. Agricultural Economics 43: 635-653. Otsuki, T., J.S. Wilson and M. Sewadeh. 2001. Saving two in a billion: quantifying the trade effect of European food safety standards on African exports. Food Policy 26: 495-514. Piggott, N.E. and T.L. Marsh. 2004. Does food safety information impact US meat demand? American Journal of Agricultural Economics 86: 154-174. Pistachio Production, Supply and Distribution online. 2016. United States Department of Agriculture-Foreign Agricultural Service (Pistachio PSD USDA-FAS). Available at: http://tinyurl.com/zfonm5s. Rapid Alert System for Food and Feed (RASFF). 2016. RASFF – Food and feed safety alerts. Available at: http://tinyurl.com/zaer5l8. Winchester, N., M.L. Rau, C. Goetz, B. Larue, T. Otsuki, K. Shutes, C. Wieck, H. Burnquist, M. Pinto de Souza and R. Nunes de Faria. 2012. The impact of regulatory heterogeneity on agri-food trade. World Economy 35: 973-993. Xiong, B. and J. Beghin. 2012. Does European aflatoxin regulation hurt groundnut exporters from Africa? European Review of Agricultural Economics 39: 589-609. Zheng, Z., S. Saghaian and M. Reed. 2012. Factors affecting the export demand for US pistachios. International Food and Agribusiness Management Review 15: 139-154.

International Food and Agribusiness Management Review

141

OPEN ACCESS International Food and Agribusiness Management Review Volume 20 Issue 1, 2017; DOI: 10.22434/IFAMR2015.0177 Received: 14 September 2015 / Accepted: 3 October 2016

Territory, environment, and healthiness in traditional food choices: insights into consumer heterogeneity RESEARCH ARTICLE Fabio Boncinelli a, Caterina Continib, Caterina Romanoa, Gabriele Scozzafavac, and Leonardo Casinib aResearch

fellow, bProfessor and cResearch associate of Agricultural Economics, Dipartimento di Gestione dei Sistemi Agrari, Alimentari e Forestali, University of Florence, Piazzale delle Cascine, 18 50144, Florence, Italy

Abstract Traditional foods are facing new market challenges tied to current trends in food habits and their determinants, such as the decline of domestic food preparation, the increased demand for convenience foods, the increasing importance of industrial food production, and the evolution of regulations on food safety. In this context our study aims at improving the knowledge of consumer segments in traditional foods market in order to develop better marketing strategies. The preferences for different credence attributes are investigated applying a latent class choice model to the extra-virgin olive oil market in Italy. Results show the existence of a marked heterogeneity of preferences, which determines the presence of both vertical and horizontal differentiation of the product. Keywords: credence attributes, latent class model, extra-virgin olive oil, consumer profiling JEL code: D12, Q13, M31 Corresponding author: fabio.boncinelli@unifi.it

ÂŠ 2016 Boncinelli et al.

143

Boncinelli et al.

Volume 20, Issue 1, 2017

1. Introduction In the postmodern age, we witness a radical evolution of the role of consumption that transcends biological needs and manifests itself in new determinants of food choices that embrace the ethical, environmental, and cultural dimensions of food (Unnevehr et al., 2010). Among these, the link between food and territory has received special attention. This is highlighted by extensive literature which has emphasized the interest of consumers for traceability (Verbeke and Roosen, 2009), typicality (Scozzafava et al., 2015), local production (Darby et al., 2008; Dentoni et al., 2009), and for all the signals and logos that permit the consumer to associate the product with the region of origin (Deselnicu et al., 2013; Hu et al., 2012). The association between food and territory is particularly strong with local and traditional foods. These kinds of food have received the attention of researchers (see Feldmann and Hamm, 2015 for an extensive review) and policy makers (DEFRA, 2005). Chambers et al. (2007) stressed that consumers perceive traditional and local foods as being of higher quality (Boyle, 2003; Fandos and Flavian, 2006), fresher (La Trobe, 2001), more nutritious, tastier, safer (Seyfang, 2004), and more sustainable (Cerjak et al., 2014; Risku-Norja et al., 2008). Guerrero et al. (2009) develop the concept of traditional food, identifying the main features to define a food as traditional: (1) a product frequently consumed over time or associated with specific celebrations or seasons; (2) the focus of strong beliefs about nutritional and sensory characteristics that should be transmitted from one generation to another; (3) the preparation and consumption is specific, in accordance with a gastronomic heritage and finally, (4) must be associated with a certain local area, region, or country. Based on the above characteristics and considering that similar beliefs determine highly homogeneous food choices (Lusk et al., 2014), we could suppose that the consumption pattern of traditional products is homogeneous. However, it is possible to find a marked supply differentiation of traditional foods on the market, especially with respect to the credence attributes, such as site of production, organic certification, and health claims. Product differentiation by credence attributes effectively seems to be the principal differentiation strategy possible for these types of product. This is particularly true considering the scarce possibility for traditional food producers to undertake a communication strategy based on different quality characteristics in the case where the law defines strict parameters, such as acidity and peroxide number for extra-virgin olive oil (EC, 1991; Thomé da Cruz and Menasche, 2014). In this context, understanding the consumers’ preferences for different credence attributes of traditional products seems crucial for the development and revival of mature markets such as traditional food products. In particular, for some of these products, this appears even more important in relation to the new market challenges tied to current trends in food habits and their determinants, such as the decline of home food preparation (Casini et al., 2013), the increased demand for convenience foods (Pieniak et al., 2009), the increasing importance of industrial food production (Kuznesof et al., 1997), and the evolution of regulations on food safety (Thomé da Cruz and Menasche, 2014). Our study therefore intends to improve the understanding of consumer preferences with respect to credence attributes for traditional products by means of: (1) consumer segmentation, identifying the dimensions most important to understand market heterogeneity; (2) consumer profiling according to socio-demographics, purchasing habits, and motivations. The connection between traditional products and consumer behaviour has been the subject of several papers, such as those on cheese in Portugal (Souza Monteiro and Ventura Lucas, 2001) and Spain (Bárcenas et al., 2001), wine in France and Spain (Sáenz-Navajas et al., 2014), and olive oil in the Mediterranean countries (Aprile et al., 2012; Chan-Halbrendt et al., 2010). However, these studies focused on several aspects of traditional food consumption and motivations without analysing the heterogeneity of consumer choices (Feldman and Hamm, 2015; Vanhonacker et al., 2010a, 2010b). In fact, few authors have concentrated directly on segmenting and profiling traditional food consumers. Vanhonacker et al. (2010b) draw a profile according International Food and Agribusiness Management Review

144

Boncinelli et al.

Volume 20, Issue 1, 2017

to socio-demographics, attitudes, lifestyle orientations, and behavioural characteristics. However, no attention has been devoted to describing the potential heterogeneity of consumer preference choices for traditional foods. Even in cases where several markets were analysed from the perspective of both segmentation and profiling, no analysis was conducted on the role of credence attributes in choice behaviours with respect to traditional products. With several traditional products facing a difficult market outlook, it seems very important to garner further knowledge of consumer segments in function of their preferences for credence attributes. This would indeed make it possible to develop better marketing strategies to calibrate the range of products or to focus communication on the most important food credence characteristics. The preferences for different credence attributes of traditional food are investigated using extra-virgin olive oil in Italy as a case study. This product fully matches the Guerrero et al. (2009) definition for traditional foods. Extra-virgin olive oil consumption in Italy is indeed habitual and strongly tied to the gastronomic tradition; moreover, there are strong beliefs about its nutritional and sensory characteristics. A review of literature on consumer preferences for olive oil reveals that important elements in the purchasing decision are represented by experience attributes like colour, taste, and smell, as well as by search and credence characteristics such as price, denomination of origin, organic certification, brand, and packaging. Overall, price is one of the most important product attributes (García et al., 2002), followed by taste, while packaging is the least important (Dekhili et al., 2011). Moreover, the studies show that the origin cues also prove to be fundamental elements in decision-making, and that the reputation of the region of origin influences the perceived overall quality (Dekhili and d’Hauteville, 2009; Delgado and Guinard, 2011; Sottomayor et al., 2010). A further determinant of preferences is represented by organic certification, which is frequently associated with health aspects (Krystallis and Chryssohoidis, 2005; Sandalidou et al., 2002; Soler et al., 2002). In general, a composite picture emerges in which the effective importance of the above attributes depends to a large extent on the consumer’s characteristics in terms of experience, awareness, and perception of the product (Fotopoulos and Krystallis, 2001; Kalogeras et al., 2009; Scarpa and Del Giudice, 2004). In this regard, a decisive factor in differentiating behaviours is represented by familiarity. There are in fact significant differences between traditional oil-producing countries and countries where oil consumption has only recently been introduced. In particular, the area of origin assumes greater importance in countries where oil is part of tradition (Espejel et al., 2009; Finardi et al., 2009; Krystallis and Ness, 2005; Manapace et al., 2011; Sottomayor et al., 2010), while in other countries, price is the most important attribute (García et al., 2002). In this paper we shall illustrate the methodological theoretical framework and then describe the choice experiment conducted on a representative sample of Italian consumers. Finally, results will be discussed, and the main managerial implications outlined.

2. Theoretical framework and methods Consumers’ preferences were analysed, employing the random utility theory (Louviere and Woodworth, 1983; McFadden, 1974; Train, 2003). The theoretical basis of this framework is in Lancaster (1966), which states that a good can be considered as a bundle of attributes, and that each one contributes to the consumer’s utility. The consumer therefore chooses a specific good according to his preference for the single features. In general, a given individual i has a set of mi mutually exclusive alternatives j that constitute his possibility set of choices. The individual will chose the alternative with greater utility. The utility assigned depends on the observable characteristics, or attributes, of the alternative itself. Therefore: U ji = U i(x ji ) + ε ji (1)

International Food and Agribusiness Management Review

145

Boncinelli et al.

Volume 20, Issue 1, 2017

where x ji is the vector of attributes relative to the alternative j and ε ji is a stochastic component of utility. The individual i will choose alternative j if the utility associated to this alternative, a function of its attributes x ji, is greater than other alternatives in the set mi. The choice experiment is based on this assumption (Carlsson et al., 2007; Lusk and Schroeder, 2004). In this study we applied a latent class model (LCM) in order to estimate the consumer preferences stated in the choice experiment. This model allows us to investigate the heterogeneity of preferences and, at the same time, obtain segmentation into groups of consumers with similar preferences (Swait, 1994; Swait and Louviere, 1993). The advantage in applying this model is the possibility to assess utility functions that are conditioned by the probability of individuals belonging to the different latent segments. In fact, the LCM is being increasingly used as an approach to account for differences in consumer preferences. This is because the traditional logit model assumes that the consumers are homogeneous (Tonsor et al., 2009). Although the random parameters logit (RPL) model incorporates preference heterogeneity into the estimation, this model is not adequate for the purpose of this paper, since RPL assumes a continuous distribution of the parameters to introduce heterogeneity without identifying discrete classes. Furthermore, the LCM outperforms traditional clustering techniques, mainly because it is based on a probability model that permits the use of inference on the results. Another benefit of the model is that it bypasses the problem of choosing linkage rules and dissimilarity measures. The choice of these elements is of great importance in cluster formation, but it is extremely hard to identify a preference criterion among them from the theoretical-economic viewpoint. Moreover, the LCM enables the calculation of statistical indicators, such as Bayesian information criteria (BIC) and Akaike information criterion (AIC), to guide the choice of the number of classes (Yang, 2006). The parameter heterogeneity is modelled across a set of latent groups or classes. Class c is latent because the individual membership is not revealed to the analyst but assigned by the model. Given a fixed number of classes c, the LCM estimates specific parameters for each class and an individual probability to belong to the classes. Thus, the utility of individual i to choose among j alternatives conditional to being in class c can be written as: Uji|c = βc′xji + εji

(2)

where Uji is the utility of alternative j to individual i; xji is the vector of attributes (certifications, health claims, site of production and price in this case); εji is the unobserved heterogeneity, and βc the class specific parameter vector. A multinomial logit model generates the choice probabilities: exp(βcxji) Prob[y = j|c] = j ∑ βcxji

(3)

j=1

The dependent variable y is represented by the choices elicited in the experiment by the respondents. The vector of parameters βc is not specific to an individual but is a class-specific parameter vector estimate. The assignment of individuals into classes is probabilistic and based on their choice. This is done in order to obtain classes where members have similar tastes and preferences. In this paper, the estimation of the LCM parameters was conducted utilising the statistical software Latent Gold Choice 4.5 (Statistical Innovation Inc., Belmont, MA, USA). Then the segments were profiled by means of chi-squared automatic interaction detection (CHAID) analysis. We have used SI-CHAID software for this purpose, integrating it with Latent Gold, which makes it possible to gather the degree of uncertainty associated with each individual’s belonging to a class.

International Food and Agribusiness Management Review

146

Boncinelli et al.

Volume 20, Issue 1, 2017

3. Choice experiment and data The characteristics of extra-virgin olive oil that were analysed in the choice experiment were certifications, health claim, site of production, and price. The certifications included protected designation of origin (PDO) and organic, which are the characteristics concerning the production process that literature indicates are the most influential in consumer choices (Aprile et al., 2012; Manapace et al., 2011). We considered four levels resulting from all of the possible combinations between the two certifications, including the absence of both (i.e. PDO, organic, PDO + organic, none). The second food attribute concerned the health claim, given the growing interest of consumers in health aspects of food (Grunert and Wills, 2007; Roosen et al., 2007; Verbeke et al., 2009). Indeed, healthiness, along with taste, represents one of the principal purchasing motivations of olive oil (Santosa et al., 2013; Santosa and Guinard, 2011). Two levels were considered, i.e. no health claims or the health claims authorised by the European Commission (EC, 2012), which is ‘Olive oil polyphenols contribute to protecting blood lipids from oxidative stress. The beneficial effect is obtained with a daily intake of 20 g of olive oil’. The site of production was included among the attributes used to analyse the importance of consumer attributes to the territory of origin. In particular, four levels for the site of production were considered: Italy, Tuscany, Apulia, and Spain. Tuscany was chosen because it represents one of the regions of Italy that is renowned for oil production, while Apulia represents the region of Italy with the greatest production in terms of quantity. The level ‘Italy’ represents a generic indication of origin, which differs from the specific indication of Tuscany or Apulia, and Spain is the other main producer in the Mediterranean. Finally, price was selected because it is one of the most important attributes in the choice of olive oils (Dekhili et al., 2011; García et al., 2002). There were six price levels, which were selected based on the distribution of prices for extra-virgin olive oils (Nielsen data; Nielsen Corporation, New York, NY, USA). The respondents were shown the choice sets without any further information that could explain the meaning of the labels used in the experiment or of the health claim. The experimental design was built by an orthogonal fractional factorial design. Each choice task contained four alternatives, as well as the no-choice option. The alternatives were presented in the form of labels that differed in their combinations of attributes. The design produced 24 choice tasks, which were divided into two blocks of 12 sets each. The information was collected in April 2013 by administering questionnaires via the internet to adult consumers who use extra-virgin olive oil. Internet surveys offer several advantages over traditional surveys. The most important concern speed and the reduced costs of collecting data (McCullough, 1998; Smith, 1997). On the other hand though, the fact that Internet users form a population that is not fully superimposable onto the general population could be a limitation. We assume that this discrepancy does not have a significant impact on the results in our case study. A company specialised in market surveys recruited the sample, which was representative of the Italian population by age, gender, and geographical area (ISTAT, 2014) (Table 1). The sample was also characterised by a high frequency of extra-virgin olive oil consumption, as 91% of respondents used it more than once per week. The analysis was based on 1000 completed questionnaires. The information collected, in addition to the socio-demographic characteristics, concerned purchasing motivations for extra-virgin olive oil on a five-point Likert scale, product purchasing behaviour, and choice experiments.

International Food and Agribusiness Management Review

147

Boncinelli et al.

Volume 20, Issue 1, 2017

Table 1. Demographic characteristics of the sample. Values (%)1

Variables Geographical area

Age

Gender Household size

Family members under 18

Occupation

Family expenditures per month

northern Italy central Italy southern Italy and islands 18-34 years 35-54 years higher than 54 male female single two members three members four members more than four none one two more than two self employed employee housewife pensioner unemployed student less than 1000 € 1000-1,999 € 2,000-2,999 € more than 3,000 €

46 (46) 20 (20) 34 (34) 28 (28) 45 (39) 27 (33) 48 (48) 52 (52) 11 23 30 27 9 67 19 12 3 16 41 11 14 9 9 40 39 13 8

1 The data between parentheses concern the Italian population in the year 2014, and were collected by the Italian National Institute of Statistics (ISTAT, 2014).

4. Results To determine the best number of classes, we used the structure of the information criteria values, i.e. the AIC and the BIC (Yang, 2006), which are shown in Table 2. The best models should minimize the two indicators. Moreover, following the suggestion of Scarpa and Thiene (2005), we also considered the significance and signs of the parameters using various segmentation hypotheses. According to these criteria we estimated a 5-class model. A preliminary analysis of the parameters across the five clusters proposed two possible interpretations. The first emerged from the evidence of a high level of non-choice in several groups and posed the question as to the motivations for rejection (Table 3). The second was based on analysing the importance of the attributes in orienting the preferences of consumers in each group. More specifically, this interpretation enabled us to consider three consumer categories. The first was represented by consumers who choose essentially on the basis of production area; the second category was made up of consumers who, in addition to production area, attribute importance to price. In the third category, finally, price was the only decisive attribute. Certifications did not represent the priority choice element in any of

International Food and Agribusiness Management Review

148

Boncinelli et al.

Volume 20, Issue 1, 2017

Table 2. Summary of latent class cluster models.1 Models

BIC(LL)

AIC(LL)

Npar

1-Cluster model 2-Cluster model 3-Cluster model 4-Cluster model 5-Cluster model 6-Cluster model 7-Cluster model

-16,050 -13,589 -12,541 -12,048 -11,789 -11,613 -11,429

32,190 27,364 25,365 24,476 24,055 23,800 23,529

32,126 27,231 25,164 24,206 23,717 23,393 23,053

13 27 41 55 69 83 97

1 LL = log likelihood; BIC = the Bayesian information criterion; AIC = Akaike information criterion; Npar = number of parameters.

Table 3. Latent class model parameter estimation.1 Attributes

Label

Site of production

Health claim Price (â&#x201A;Ź)

No-choice

Levels

Clusters

none PDO+Organic2 organic PDO Spain Italy Apulia Tuscany none claim 3 6 9 12 15 18 choice no-choice

Quality seekers

Pragmatics

Price sensitive

Hard-toplease

Nochoosers

0 1.344** 0.635** 0.715** 0 1.422** 1.863** 1.944** 0 0.370** 0 0.742** 1.012** 0.787** 0.377** -0.250** 0 -1.148**

0 1.011** 0.625** 0.529** 0 2.022** 2.005** 2.298** 0 0.205 0 -0.208 -1.183** -2.422** -4.728** -5.561** 0 -1.574**

0 -0.716** -1.796** -0.531** 0 2.135** 2.855** 1.829** 0 0.158 0 -2.097** -5.896** -5.864** -9.627** -11.992* 0 4.083**

0 1.146** 0.636* 0.77** 0 2.789** 2.959** 2.989** 0 0.59** 0 0.819** -0.686** -2.079** -4.434** -4.741** 0 2.213**

0 -0.818** -0.453* -0.046 0 2.103** 1.921** 2.113** 0 0.206 0 -0.031 -0.201 -0.839** -0.921** -0.926** 0 3.916**

1 ** 2

and * denote significance at the 1 and 5% level, respectively. PDO = protected designation of origin.

the cases. The PDO and Organic labels influenced the choices of several types of consumers, while health claims played a more marginal role. For a better understanding of the characteristics of the various segments that the LCM identified, we decided to integrate the information with the analysis of the socio-demographics, the purchasing channels, and the motivational variables. Table 4 reports the results of the CHAID analysis which underline that the market segments significantly differ as per three socio-demographic variables (i.e. age, occupation and family expenditure). As far as motivational variables are concerned, two categories were identified â&#x20AC;&#x201C; personal and relational motivations. As for personal motivations, culinary habits and taste did not prove to be significantly different among the groups. This non-predictable result shows that a marked heterogeneity of choice is accompanied International Food and Agribusiness Management Review

149

Boncinelli et al.

Volume 20, Issue 1, 2017

Table 4. Levels of significance of the variables used for profiling. Categories

Variables

Relational motivations familiarity with the producer information campaigns physician’s advice advertising Personal motivations part of the Mediterranean diet healthier than other condiments culinary habits taste Purchasing channels direct sale shops specialized in the sale of quality products grocery stores supermarket farmers’ market discount store Socio-demographics family expenditures per month occupation1 age number of members number of members under 18 gender1 geographical area1 1

Log-likelihood ratio Chi square

P-value

148.98 79.65 45.86 35.05 14.03 13.36 0.00 0.00 129.10 92.40 64.35 61.97 47.47 54.38 37.44 37.00 14.07 10.47 0.00 0.00 0.00

12 8 8 4 4 4 0 0 12 4 8 8 4 8 8 8 4 4 0 0 0

<0.001 <0.001 <0.001 <0.001 0.029 0.038 1 1 <0.001 <0.001 <0.001 <0.001 <0.001 <0.001 <0.001 0.011 0.014 0.44 1 1 1

These variables are nominal. All the others are ordinal.

by some homogeneity of declared choice motivations. Instead, the relational motivations proved to be significantly different among the various segments. This evidence shows that in these contexts of choice, the communication variables tied to personal relationships have a great influence on purchasing behaviour. Figure 1 reports the distributions of the statistically significant variables used to profile the 5 clusters identified. In accordance with the LCM results, the five clusters were labelled as follows: quality seekers (32%), pragmatic consumers (16%), price-sensitive consumers (16%), hard-to-please consumers (19%), and nonchoosers (18%). It is worth noting that not only is the last group characterised by a high percentage of nonchoice, but so are the price-sensitive and hard-to-please consumers as well. The quality seekers are the group that attributes great importance to certifications, considering them to be an element of product quality. In particular, the importance of the PDO and Organic labels emerges and further increases when these labels are present at the same time. However, the group’s decisive attribute is represented by the area of origin, meaning both its region of production and Italian origin. As far as price is concerned, this group shows a preference for the € 6.00 to 12.00 range, which reflects the market values that are associated with products of a fair quality level. This result indicates a perception of price as a quality cue. Therefore, the inclination towards quality is a distinctive feature of the segment, which is also confirmed by the fact that the preferred distribution channels are grocery stores, farmers’ markets and shops, which specialise in the sale of quality products. The search for quality is also determined by the attention to health and quality seekers’ sensitivity towards campaigns that promote a healthy diet and the advice of physicians (Figure 1). Pragmatic consumers have inclinations that are similar to those of quality seekers, as far as area of origin and certifications are concerned. However, the discriminating factor between the two groups is the attitude International Food and Agribusiness Management Review

150

Boncinelli et al.

Volume 20, Issue 1, 2017

Pragmatics Price sensitive Hard-to-please No choosers 0%

Min Quality seekers Pragmatics Price sensitive Hard-to-please No choosers

20% 40% 60% 80% 100%

Quality seekers

Pragmatics

Advertising

Quality seekers

Price sensitive Hard-to-please

Hard-to-please

20% 40% 60% 80% 100%

Quality seekers Pragmatics Price sensitive Hard-to-please No choosers 0%

Price sensitive

No choosers

No choosers 0%

Part of the Mediterranean diet

Information campaigns

Quality seekers

20% 40% 60% 80% 100%

Healthier than other condiments

Physicianâ&#x20AC;&#x2122;s advice

Familiarity with the producer

Max

Quality seekers Pragmatics Price sensitive Hard-to-please No choosers

Figure 1. Segment profile defined by means of relational and personal motivations. towards price. The sign and coefficients of price level in fact show an inclination towards choosing products with lower prices. This attitude seems to decisively guide the behaviour of the pragmatic consumer, insofar as the strength of price overshadows the product characteristics that are associated with quality. Consequently, the relative importance of certifications on this groupâ&#x20AC;&#x2122;s choices was lower than that of the quality seekers. This behaviour is also reflected in the purchasing channels. Indeed, the respondents who buy the product exclusively or prevalently in supermarkets or discount stores have a greater probability of belonging to this group (Figure 2). Moving on to the analysis of the segments with the highest frequency of non-choice, the price sensitive consumers are characterised by non-choice motivations tied to price, in the sense that they exclusively purchase products in the lower price bracket and mainly from discount stores. It is noteworthy that, for these consumers, the parameters of labels are negative and significant. This implies that, for the price sensitive, the PDO and Organic certifications are associated with a negative utility. This result can be explained by the fact that, for these consumers, certifications combined with low prices are not only uninteresting but can even prove incoherent with top bracket prices. Therefore, they should be considered as negative indicators of quality. The predominance of the price attribute is responsible for the fact that this cluster was not receptive to any type of external information, whether of a health-oriented or commercial type (Figure 1 and Figure 2). The hard-to-please consumer group is characterised by a purchasing behaviour that requires high quality levels (as shown by the significance of the parameters of origin and certification). At the same time, it showed International Food and Agribusiness Management Review

151

Volume 20, Issue 1, 2017

Direct sale

Quality seekers Pragmatics Price sensitive Hard-to-please No choosers

Min Quality seekers

Price sensitive Hard-to-please No choosers 0%

20% 40% 60% 80% 100%

Quality seekers

Pragmatics

Price sensitive Hard-to-please

Hard-to-please

20% 40% 60% 80% 100%

Quality seekers Discount stores

Quality seekers Pragmatics Price sensitive Hard-to-please

Pragmatics Price sensitive Hard-to-please No choosers

No choosers 0%

Price sensitive

No choosers

Farmersâ&#x20AC;&#x2122; market

Pragmatics

20% 40% 60% 80% 100%

Supermarket

Grocery stores

Max Shops specialized in the sale of quality products

Boncinelli et al.

20% 40% 60% 80% 100%

Figure 2. Segment profile defined by means of purchasing channels. a growing utility only at prices no higher than â&#x201A;Ź 6.00. Beyond this level, the product quality characteristics seemed to assume a lesser importance, compared to the increases in price. This behaviour consequently determines the difficulty of choice in many choice tasks, which corresponds to the high rates of non-choice that were surveyed. Finally, a particularly interesting group was that of the non-choosers. This group is characterised by a very high non-choice frequency, as shown by the value of the non-choice parameters in both absolute and relative terms. The explanation of this behaviour can probably be found in the prevalent purchasing habits of the group, which proved inclined to turn directly to the producer, avoiding the traditional channels. For this reason, faced with a choice experiment between alternatives of packaged oils, the nonchoosers refused to choose. The analysis of the motivational variables confirmed this hypothesis and pointed out that familiarity with the producer is particularly important for the non-choosers (Figure 1). In this case, direct contact with the producer becomes the principal guarantee of quality, taking the place of other signals such as certification. The five segments were profiled according to the three socio-demographic characteristics that proved to be discriminant between the classes: age, occupation, and monthly family expenditure (Figure 3). The analysis shows a clear prevalence of older consumers among the non-choosers, while younger consumers are predominant among the quality seekers. As far as the other segments are concerned though, the differences with respect to age are marginal. Profiling by occupation is coherent with profiling by age, as pensioners predominate in the group of non-choosers. Moreover, it emerges that the inactive (i.e. housewives, students) and the unemployed are concentrated in the hard-to-please segment. The employed are mostly distributed International Food and Agribusiness Management Review

152

Boncinelli et al.

Volume 20, Issue 1, 2017 Family expenditures per month Quality seekers

Less than 1000 € 1000-1,999 € 2,000-2,999 € More than 3,000 €

Pragmatics Price sensitive Hard-to-please No choosers 0%

20%

40%

60%

80%

100%

Occupation Quality seekers Pragmatics Price sensitive Hard-to-please No choosers

Self-employed Employee Housewife Pensioner Unemployed Student

20%

40%

60%

80%

100%

Age Quality seekers Pragmatics

18-34 35-54 >54

Price sensitive Hard-to-please No choosers 0%

20%

40%

60%

80%

100%

Figure 3. Segment profile defined by means of socio-demographic variables. among the quality seekers and the pragmatics. Moving on to consider family expenditure, the respondents with higher spending levels fall in the classes that show more attention to quality. In fact, the weight of individuals with a family spending higher than 2,000 euros per month is greater in the classes of the quality seekers, while among the price-sensitive consumers and the non-choosers, the percentage with a family expenditure under 1000 euros per month is predominant.

5. Final considerations The analysis of purchasing behaviours for traditional foods indicates the existence of a marked heterogeneity of preferences even where these foods are part of the traditional diet. For several groups of consumers, the product takes on features typical of a vertical differentiation (that is to say, a single qualitative ranking exists), while for other groups, a horizontal differentiation can be identified (the consumers have different concepts of quality). In fact, our study has brought to light various market segments representative of markedly differentiated choice behaviours, though presenting several shared characteristics. The site of production shows a significant and positive impact on consumer preferences, thus confirming previous studies on traditional foods (Chan-Halbrendt et al., 2010; Feldmann and Hamm, 2015). This represents an element of homogeneity of choices between clusters, though the relative importance of this attribute varies considerably between segments. Another element of homogeneity is the limited importance of the health claim. This finding apparently contrasts with the literature on health claims, which states that, for various products, health claims have a distinctly positive impact on choices (Grunert and Wills, 2007; Roosen et al., 2007; Verbeke et al., 2009). This can be explained by the fact that consumers of traditional foods have a large amount of knowledge about the outcomes attained from products due to repeated consumption in time. The information that the claim guarantees is therefore probably already known to the consumer, who thus attributes less importance to certification. Consequently, a more efficient use of health claims to promote traditional food should provide more specific information, that is to say information with a greater selectivity with respect to the different quality levels in nutraceutical terms. In the case of extra-virgin olive

International Food and Agribusiness Management Review

153

Boncinelli et al.

Volume 20, Issue 1, 2017

oil, this goal could be pursued by defining more discriminating thresholds and parameters than those that European regulations currently provide. Among the attributes that most influence the heterogeneity of choices, price and PDO and Organic certifications have emerged. Price appears to take on contrasting roles. For some segments, it constitutes a signal of value that determines the consumer’s preference, thereby resulting in a preference for products with at least a certain price level. For other segments, low prices represent the main factor of preference in any event. In particular, in our study, a relevant share of consumers, almost one-third, is driven by price, and about half of these consumers assume the same decision-making process that they use with a commodity. For this segment, the marketing strategies that seem more efficient would therefore be those tied to the various types of promotional sales. Another element that emerges from our study is the fact that certification can assume a negative utility for consumers oriented exclusively by price. In fact, in this case, the signal of certifications associated with low prices is perceived as incoherent information that increases the uncertainty about the product’s effective properties. This behaviour is coherent with the ‘More is less’ preference reversal phenomenon. According to this theory, in the presence of uncertainty, consumers perceive an additional positive attribute as a cue that the rest of the good is not worth much (List, 2002). The analysis of the profiles of consumers belonging to the different segments identified has pointed out the important role of purchasing channels, while among the socio-demographic variables used, age, profession, and average family expenditure have proved particularly discriminating. All three of these are reasonably correlated. Age proves to be the variable that can most differentiate behaviour with respect to a traditional product, as the ties with traditions for the younger generations can be attenuated compared to the past, while the influences of food habit trends on the overall level can prove stronger. As far as distribution channels are concerned, our study has pointed out a strong relation between preferences and where the product is purchased. For olive oil, the direct relation with the producer proved to be the most important, especially for older people. This connection between tradition, preferences, and purchasing channels that emerged in our study can probably be extended also to many other traditional products. Hence the choice of place of purchase forms an integral part of the almost ritual relationship with the product. These results can also carry interesting managerial implications. The different behaviour of the new generations in choosing purchasing channels allows us to glimpse the opportunity for new means to sell even traditional products. E-commerce, in particular, could constitute a means of bringing the producer and the consumer into a direct relationship in ways more consistent with new lifestyles. Finally, our results underline the existence of numerically consistent segments of consumers characterised by the same scale of preferences (vertical differentiation) that can currently be explained by a few quality cues. Particularly interesting in this ambit appears to be the possibility to increase the available information on quality, so as to enable the consumer to improve his choice process and thus provide more tools for product differentiation. This could be the case of the production techniques or the product’s objective characteristics, such as the content of specific elements associated with taste, genuineness, or nutritional value. Future studies could aim precisely at defining these new signals of value, possibly in a multidisciplinary context capable of developing the aspects tied to product quality in a holistic perspective. The limits of this paper can be explained by the fact that our study concerned the specific case of extra-virgin olive oil in Italy. Extending analysis to other case studies could contribute to enrich the picture, strengthening the generalisation of the results.

International Food and Agribusiness Management Review

154

Boncinelli et al.

Volume 20, Issue 1, 2017

References Aprile, M., V. Caputo and R. Nayga. 2012. Consumers’ valuation of food quality labels: the case of the European geographic indication and organic farming labels. International Journal of Consumer Studies 36: 158-165. Bárcenas, P., R. Pérez de San Román, F. Pérez Elortondo and M. Albisu. 2001. Consumer preference structures for traditional Spanish cheeses and their relationship with sensory properties. Food Quality and Preference 12: 269-279. Boyle, D. 2003. Authenticity: brands, fakes, spin and the lust for real life. Flamingo, London, UK. Carlsson, F., P. Frykblom and C. Lagerkvist. 2007. Preferences with and without prices – does the price attribute affect behavior in stated preference surveys? Environmental and Resource Economics 38: 155-164. Casini, L., C. Contini, E. Marone and C. Romano. 2013. Food habits. Changes among young Italians in the last 10 years. Appetite 68: 21-29. Cerjak, M., R. Haas, F. Brunner and M. Tomić. 2014. What motivates consumers to buy traditional food products? Evidence from Croatia and Austria using word association and laddering interviews. British Food Journal 116: 1726-1747. Chambers, S., A. Lobb, L. Butler, K. Harvey and B. Traill. 2007. Local, national and imported foods: a qualitative study. Appetite 49: 208-213. Chan-Halbrendt, C., E. Zhllima, G. Sisior, D. Imami and L. Leonetti. 2010. Consumer preferences for olive oil in Tirana, Albania. International Food and Agribusiness Management Review 13: 55-74. Darby, K., M.T. Batte, S. Ernst and B. Roe. 2008. Decomposing local: a conjoint analysis of locally produced foods. American Journal of Agricultural Economics 90: 476-486. Department for Environment, Food and Rural Affairs (DEFRA). 2005. The validity of food miles as an indicator of sustainable development. Defra, London, UK. Available at: http://tinyurl.com/hxqtwlu. Dekhili, S. and F. d’Hauteville. 2009. Effect of the region of origin on the perceived quality of olive oil: an experimental approach using a control group. Food Quality and Preference 20: 525-532. Dekhili, S., L. Sirieix and E. Cohen. 2011. How consumers choose olive oil: the importance of origin cues. Food Quality and Preference 22: 757-762. Delgado, C. and J. Guinard. 2011. How do consumer hedonic ratings for extra virgin olive oil relate to quality ratings by experts and descriptive analysis ratings? Food Quality and Preference 22: 213-225. Dentoni, D., G. Tonsor, R. Calantone and H. Peterson. 2009. The direct and indirect effects of ‘Locally Grown’ on consumers’ attitudes towards agri-food products. Agricultural & Resource Economics Review 38: 384-396. Deselnicu, O.C., M. Costanigro, D.M. Souza-Monteiro and D.T. McFadden. 2013. A meta analysis of geographical indication food valuation studies: what drives the premium for origin-based labels? Journal of Agricultural and Resource Economics 38: 204-219. Espejel, C., C. Fandos and C. Flavián. 2009. The influence of consumer degree of knowledge on consumer behavior: the case of Spanish olive oil. Journal of Food Products Marketing 15: 15-37. European Commission (EC). 1991. Commission Regulation (EEC) No 2568/91 of 11 July 1991 on the characteristics of olive oil and olive-residue oil and on the relevant methods of analysis. Official Journal of the European Union L 248: 1-83. European Commission (EC). 2012. Commission Regulation (EU) No. 432/2012 of 16 May 2012 establishing a list of permitted health claims made on foods, other than those referring to the reduction of disease risk and to children’s development and health. Official Journal of the European Union L 136: 1-40. Fandos, C. and C. Flavian. 2006. Intrinsic and extrinsic quality attributes, loyalty and buying intention: an analysis for a PDO product. British Food Journal 108: 646-662. Feldmann, C. and U. Hamm. 2015. Consumers’ perceptions and preferences for local food: a review. Food Quality and Preference 40: 152-164. Finardi, C., C. Giacomini, D. Menozzi and C. Mora. 2009. Consumer preferences for country-of-origin and health claim labelling of extra-virgin olive-oil, in a resilient European food industry and food chain in a challenging world. Available at: http://tinyurl.com/js8rqno. International Food and Agribusiness Management Review

155

Boncinelli et al.

Volume 20, Issue 1, 2017

Fotopoulos, C. and A. Krystallis. 2001. Are quality labels a real marketing advantage? A conjoint application on Greek PDO protected olive oil. Journal of International Food and Agribusiness Marketing 12: 1-22. García, M., Z. Aragonés and N. Poole. 2002. A repositioning strategy for olive oil in the UK market. Agribusiness 18:163-180. Grunert, K.G. and J.M. Wills. 2007. A review of European research on consumer response to nutrition information on food labels. Journal of Public Health 15: 385-399. Guerrero, L., M.D. Guàrdia, J. Xicola, W. Verbeke, F. Vanhonacker, S. Zakowska-Biemans, M Sajdakowska, C. Sulmont-Rossé, S. Issanchou, M. Contel, M.L. Scalvedi, B.S. Granli and M. Hersleth. 2009. Consumer-driven definition of traditional food products and innovation in traditional foods. A qualitative cross-cultural study. Appetite 52: 345-354. Hu, W., M.T. Batte, T. Woods and S. Ernst. 2012. Consumer preferences for local production and other value-added label claims for a processed food product. European Review of Agricultural Economics 39: 489-510. Istituto nazionale di statistica (ISTAT). 2014. Italian warehouse of statistics. Available at: http://dati.istat.it. Kalogeras, N., S. Valchovska, G. Baourakis and P. Kalaitzis. 2009. Dutch consumers’ willingness to pay for organic olive oil. Journal of International Food and Agribusiness Marketing 21: 286-311. Krystallis, A. and G. Chryssohoidis. 2005. Consumers’ willingness to pay for organic food: Factors that affect it and variation per organic product type. British Food Journal 107: 320-343. Krystallis, A. and M. Ness. 2005. Consumer preferences for quality foods from a South European perspective: a conjoint analysis implementation on Greek olive oil. International Food and Agribusiness Management Review 8: 62-91. Kuznesof, S., A. Tregear and A. Moxey. 1997. Regional foods: a consumer perspective. British Food Journal 99: 199-206. La Trobe, H. 2001. Farmers’ markets: consuming local rural produce. International Journal of Consumer Studies 25: 181-192. Lancaster, K. 1966. A new approach to consumer theory. Journal of Political Economy 74: 132-157. List, J.A. 2002. Preference reversals of a different kind: the ‘more is less’ phenomenon. American Economic Review 92: 1636-1643. Louviere, J.J. and G. Woodworth. 1983. Design and analysis of simulated consumer choice or allocation experiments: an approach based on aggregate data. Journal of Marketing Research 20: 350-367. Lusk, J.L. and T.C. Schroeder. 2004. Are choice experiments incentive compatible? A test with quality differentiated beef steaks. American Journal of Agricultural Economics 86: 467-482. Lusk, J.L., T.C. Schroeder and G.T. Tonsor. 2014. Distinguishing beliefs from preferences in food choice. European Review of Agricultural Economics 41: 627-655. Manapace, L., G. Colson, C. Grebitus and M. Facendola. 2011. Consumers’ preferences for geographical origin labels: evidence from the Canadian olive oil market. European Review of Agricultural Economics 38: 193-212. McCullough, D. 1998. Web-based market research, the dawning of a new era. Direct Marketing 61: 36-39. McFadden, D. 1974. Conditional logit analysis of qualitative choice behaviour. In Frontiers in Econometrics, edited by P. Zarembka. Academic Press, New York, USA, pp.105-142. Pieniak, Z., W. Verbeke, F. Vanhonacker, L. Guerrero and M. Hersleth. 2009. Association between traditional food consumption and motives for food choice in six European countries. Appetite 53: 101-108. Risku-Norja, H., R. Hietala, H. Virtanen, H. Ketomaki and J. Helenius. 2008. Localisation of primary food production in Finland: Production potential and environmental impacts of food consumption patterns. Agricultural and Food Science 17: 127-145. Roosen, J., S. Marette, S. Blanchemanche and P. Verger. 2007. The effect of product health information on liking and choice. Food Quality and Preference 18: 759-770. Sáenz-Navajas, M., J. Ballester, D. Peyron and D. Valentin. 2014. Extrinsic attributes responsible for red wine quality perception: a cross-cultural study between France and Spain. Food Quality and Preference 35: 70-85. Sandalidou, E., G. Baourkis and Y. Siskos. 2002. Customers’ perspectives on the quality of organic olive oil in Greece: a satisfaction evaluation approach. British Food Journal 104: 391-406. International Food and Agribusiness Management Review

156

Boncinelli et al.

Volume 20, Issue 1, 2017

Santosa, M., E.J. Clow, N. Sturzenberger and J.X. Guinard. 2013. Knowledge, beliefs, habits and attitudes of California consumers regarding extra virgin olive oil. Food Research International 54: 2104-2111. Santosa, M. and J.X. Guinard. 2011. Means-end chains analysis of extra virgin olive oil purchase and consumption behaviour. Food Quality and Preference 22: 304-316. Scarpa, R. and T. Del Giudice. 2004. Market segmentation via mixed logit: extra-virgin olive oil in urban Italy. Journal of Agricultural and Food Industrial Organization 2: 1-18. Scarpa, R. and M. Thiene. 2005. Destination choice models for rock-climbing in the North-Eastern Alps: a latent-class approach based on intensity of participation. Land Economics 81: 426-444. Scozzafava, G., F. Boncinelli, C. Contini, C. Romano, F. Gerini and L. Casini. 2015. Typical vine or international taste: wine consumers’ dilemma between beliefs and preferences. Recent patents on food, nutrition and agriculture 8: 31-38. Seyfang, G. 2004. Consuming values and contested cultures: a critical analysis of the UK strategy for sustainable consumption and production. Review of Social Economy 62: 323-33. Smith, C.B. 1997. Casting the net: surveying an internet population. Journal of Computer Mediated Communication 3: 43-49. Soler, F., J.M. Gil and M. Sánchez. 2002. Consumers’ acceptability of organic food in Spain: results from an experimental auction market. British Food Journal 104: 670-687. Sottomayor, M.J., D.M. Souza Monteiro and M.S. Teixeira. 2010. Valuing nested names in the Portuguese olive oil market: an exploratory study. Available at: http://tinyurl.com/qzld7rb. Souza Monteiro, D.M. and M.R. Ventura Lucas. 2001. Conjoint measurement of preferences for traditional cheeses in Lisbon. British Food Journal 103: 414-424. Swait, J.R. 1994. A structural equation model of latent segmentation and product choice for cross-sectional revealed preference choice data. Journal of Retailing and Consumer Services 1: 77-89. Swait, J.R. and J.J. Louviere. 1993. The role of the scale parameter in the estimation and comparison of multinomial logit models. Journal of Marketing Research 30: 305-314. Thomé da Cruz, F. and R. Menasche. 2014. Tradition and diversity jeopardised by food safety regulations? The Serrano Cheese case, Campos de Cima da Serra region, Brazil. Food Policy 45: 116-124. Tonsor, G.T., N. Olynk and C. Wolf. 2009. Consumer preferences for animal welfare attributes: the case of gestation crates. Journal of Agricultural and Applied Economics 41: 713-730. Train, K. 2003. Discrete Choice Methods with Simulation. Cambridge University Press, New York, USA. Unnevehr, L., J. Eales, H. Jensen, J. Lusk, J. McCluskey and J. Kinsey. 2010. Food and consumer economics. American Journal of Agricultural Economics 92: 506-521. Vanhonacker, F., V. Lengard, M. Hersleth and W. Verbeke. 2010. Profiling European traditional food consumers. British Food Journal 112: 871-886. Vanhonacker, F., W. Verbeke, L. Guerrero, A. Claret, M. Contel, L. Scalvedi, S. Żakowska-Biemans, K. Gutkowska, C. Sulmont-Rossé, J. Raude, B. Granli and M. Hersleth. 2010. How European consumers define the concept of traditional food: evidence from a survey in six countries. Agribusiness 26: 453-476. Verbeke, W. and J. Roosen. 2009. Market differentiation potential of origin, quality and traceability labelling. Estey Centre Journal of International Law and Trade Policy 10: 20-35. Verbeke, W., J. Scholderer and L. Lähteenmäki. 2009. Consumer appeal of nutrition and health claims in three existing product concepts. Appetite 52: 684-692. Yang, C.C. 2006. Evaluating latent class analysis models in qualitative phenotype identification. Computational Statistics and Data Analysis 50: 1090-1104.

International Food and Agribusiness Management Review

157

OPEN ACCESS International Food and Agribusiness Management Review Volume 20 Issue 1, 2017; DOI: 10.22434/IFAMR2016.0002 Received: 1 April 2016/ Accepted: 3 November 2016

Investigating the impact of maximum residue limit standards on the vegetable trade in Japan RESEARCH ARTICLE Jong Woo Choia and Chengyan Yue aResearch

Fellow, Korea Rural Economic Institute, 601 Bitgaram-ro, Naju-si, Jeollanam-do 58217, Republic of Korea

bAssociate

Professor, Department of Applied Economics and Department of Horticultural Science, University of Minnesota, 458 Alderman Hall, 1970 Folwell Ave, St. Paul, MN 55108, USA

Abstract Countries have become increasingly concerned about the safety of their food. Many countries have imposed standards on both domestically produced and imported food. In particular, countries have implemented regulations to control the quantity and quality of vegetable imports. Maximum Residue Limit (MRL) standards are one of the main restrictions adopted by numerous countries. Japan has one of the strictest MRL standards in the world. This study builds on previous studies to explore the impact of MRL standards on Japanese vegetable imports. Gravity models are used to analyze how MRL standards influence the Japanese imports of different types of vegetables (fruit vegetables, leafy vegetables, bulb vegetable, and root vegetables). The results reveal that the trade impacts of MRL standards are different for different types of vegetables, with the most significant impact on imports of leafy and fruit vegetables and the least significant impact on imports of bulb vegetables. Keywords: food safety, Japan, MRL, pesticides, vegetable JEL code: Q17, F13, Q18 Corresponding author: yuechy@umn.edu

ÂŠ 2017 Choi and Yue

159

Choi and Yue

Volume 20, Issue 1, 2017

1. Introduction With outbreaks of a myriad of food-borne diseases, food safety has become a global concern. Consumers are becoming more aware of eating safer and healthier food that is domestically produced and imported from other countries. In an effort to import safe food, governments have sought ways to set up trade restrictions to ensure that imported food is safe. These restrictions include Maximum Residue Limit (MRL) standards. MRL standards set the maximum acceptable level of a specific pesticide residue on agricultural products that may enter the agricultural markets. According to the Food and Agriculture Organization and the World Health Organization (2006), strict MRL standards naturally reflect how serious nations are about their food quality, and they can help improve food safety. However, when the MRL standards are too strict, they may act as technical barriers to trade because governments may use the MRL standards to limit imports and protect the domestic food industry (Martinez and Thornsbury, 2010). These strict restrictions can generate considerable welfare losses for domestic consumers and merchandise losses for food exporters. Previous research showed that MRL standards affect the trade of agricultural products between countries. Achterbosch et al. (2009) analyzed the impact of MRL standards on the trade of fresh fruit between Chile and the European Union and found that more stringent MRL standards negatively affect trade volume. Similar results were also confirmed by Bao and Qiu (2009), who examined China’s trade barriers to imported agricultural products and processed food in China. Their papers used the gravity model to compare MRL standards between nations and how these differences affected a country’s trade. They argued that the MRL standards’ impact on trade is different for developed and developing countries. Wei et al. (2012) and Yue et al. (2010) analyzed the MRL standards’ impact on China’s honey exports and tea exports. These studies have confirmed that the trade volume is negatively affected by the stringency of MRL standards. Wilson and Otsuki (2004) reported how governments’ pesticide regulations were influenced by MRL standards. They have also confirmed that stricter MRL standards affect the trade volume negatively. Otsuki, Wilson and Sewadeh (2001a,b) also analyzed the impact of Aflatoxin standards on trade between nations. These studies concluded that Aflatoxin standards negatively affect the trade value of groundnuts. Liu and Yue (2012) extended the previous studies to analyze how the similarity of MRL regulations between nations affected substitution elasticity of domestic and imported goods by using a variable elasticity of substitution model. They found that developing countries could expand their exports to developed countries by setting stricter MRL standards. Japan is among the top countries that import a large quantity of fruits and vegetables. Japan has also applied the most stringent MRL standards in the world. The MRL standards apply to domestic vegetables as well as imported vegetables. In short, if vegetables do not meet the standards, they are not allowed to enter the market or cannot be imported in the first place. Using similarity indexes, Liu and Yue (2012) and Drogue and DeMaria (2010) compared the MRL standards’ stringencies between countries. Their studies confirmed Japan has the strictest MRL standards for each vegetable. To export vegetables to Japan, exporters must pay close attention to Japan’s strict MRL standards even though their own MRL standards are far less strict. For exporters whose own countries’ MRL standards are not as strict as Japan’s, meeting those standards means adding relatively high costs to their products. However, when the costs are too high, some exporters cannot afford to export their products to Japan because the profit margin is too narrow. In these cases, Japan’s strict MRL standards become technical barriers to trade for these exporters. Both foreign and domestic vegetable producers striving to meet Japan’s high MRL standards have to bear MRL compliance costs, which lead to higher production costs and thus decrease the profit margins. However, for countries where the input costs are lower than Japan (for example, Vietnam and Philippine have lower labor costs; Australia and Canada have lower land costs), they can comply with the MRL standards with much lower costs. The lower compliance costs can help keep vegetable price lower than the vegetables produced in Japan.

International Food and Agribusiness Management Review

160

Choi and Yue

Volume 20, Issue 1, 2017

Japan’s strict restrictions and regulations on agricultural product imports have led to some trade disputes in the World Trade Organization. For example, in 1997, the United States requested consultations with Japan to discuss Japan’s prohibition of imports of certain agricultural products such as apricots, apples, and walnuts. The United States complained that Japan prohibits importing each variety of a product that requires quarantine treatment until the quarantine treatment has been tested for that variety, even if the treatment was proved as effective for other varieties of the same product (WTO, 2013a). In the same year, the European Commission contended that certain measures affecting imports of pork and its processed products imposed by Japan violated Japan’s obligations under certain articles(Articles I, X:3 and XIII) of the GATT 1994 and nullified or impaired benefits accruing to the Commission (WTO, 2013b). In 2002, the United States complained to Japan about its measures including prohibition of imported apples from orchards in which any fire blight was detected. Japan required that export orchards be inspected three times annually for the presence of fire blight and other problems. The United States claimed that these measures were inconsistent with Japan’s obligations (WTO, 2013c). Despite the fact that countries have complained about Japan’s strict restrictions and regulations of agricultural imports, some of which led to trade disputes, Japan has persisted in maintaining the strictest MRL standards. Although Japan has the largest number of pesticide restrictions, the strictest MRL standards, and restrictions that cover almost all vegetables, no study has investigated how Japan’s strict MRL standards could affect vegetable imports to Japan. This research aims to fill this gap in the literature. First, little work has examined how the MRL standards could affect the trade of different types of food. In this study, we investigate how Japan’s MRL standards affect vegetable imports. Second, vegetables are categorized into different types and we analyze how the MRL standards might affect the imports of these different types of vegetables (i.e. leafy vegetables, fruit vegetables, bulb vegetables, and root vegetables) in different ways. The study will help determine whether the stringency of MRL standards impedes Japanese vegetable imports, and if so, to what extent the stringency of MRL standards has impeded vegetable imports. Third, we examine whether the degree to which MRL standards impede the different types of vegetables imports differently can be measured. Possible policy implications from this study are whether or not the Japanese government should make their MRL standards less strict. If so, this then demands an answer to the question of whether or not the Japanese government should adjust their MRL standards for different types of vegetables to the same extent. The rest of this paper is organized as follows: the first section is the materials and methods section that introduces the similarity index, the description of the gravity model we used, and the vegetables chosen for our analysis, a detailed description of the data used. The second section covers the results and discussion. Finally, the last section concludes the paper with a discussion of important policy implications and possible future research topics.

2. Materials and methods In this section, the similarity index is introduced firstly to better understand differences between countries’ MRL standards, and the gravity model is presented. The vegetables used in our analysis, how to categorize the vegetables, and the data description are presented in great detail. The similarity index Numerous research studies have attempted to understand the discrepancy of MRL standards among countries. They have created indexes to explain the extent to which the MRL standards are different between nations. Achterbosch et al. (2009) used an index of regulatory heterogeneity to analyze the impact of differences in MRL standards on the fresh fruit trade between Chile and the European Union. They studied how the relative differences in MRL regulations affected trade flows between importing and exporting countries. Their results showed that more similar MRL regulations with the European Union standards would increase exports of Chilean fruit. Winchester et al. (2012) also used the heterogeneity index of MRL standards for their study of the agricultural food trade between the European Union and nine other countries. They found International Food and Agribusiness Management Review

161

Choi and Yue

Volume 20, Issue 1, 2017

that imposing stricter MRL standards could reduce the trade volume. However, this index falls short of capturing the MRL standards’ differences if the MRL standards’ levels are similar but the number of MRL standards is different. In addition, Drogue and DeMaria (2010) used Pearson’s coefficient correlation index to capture the difference of MRL standards for apples and the distance between countries. The most widely used index is Jaffe’s (1986) similarity index. The strength of the index is that it can capture the ratio between a country’s MRL standard and the highest MRL standards. Some researchers have adopted the similarity index to compare the similarities between certain regulations across nations or regions. For example, Anderson (2009, 2010) used the index to compare the regional regulation similarity between Australia and other countries. By using the similarity index, he analyzed the degree to which the similarity of regulations affected the trade of wine. In the current study, Jaffe’s similarity index is adopted to understand the extent to which the MRL standards are different between Japan and other nations. The similarity index is defined as follows: ∑Nn=1 rin rjn MRLij = N (1) (∑ n=1 rin2)0.5 (∑Nn=1 rjn2)0.5 where rin is the ratio of the MRL level in country i to the highest MRL level of pesticide n. N is the total number of pesticides regulated for a product. The index is symmetric and varies from 0 to 1. This index is 1 if the MRL standards are identical between the importing and exporting countries, and is 0 if they are totally different. This study uses the index as a measure of the extent to which exporters’ MRL standards are similar to those of Japan. The gravity model To estimate the trade impacts of MRL stringency on Japan’s vegetable market, this study adopted the most widely used model: the gravity model. The gravity model has been widely used to estimate how MRL standards affect the food trade. Tinbergen (1962) applied the gravity model to understand bilateral trade flows, in general. Since Tinbergen’s study, the gravity model has become one of the main methods for analyzing how certain factors affect international trade. Several papers have investigated the appropriate practical forms for the gravity model. Silva and Tenreyro (2006) took the log of dependent and independent variable in a gravity model to verify unbiased estimator under heteroscedasticity exists. Also, Carrère (2006) used a gravity model to find out the appropriate number of dummy variables to identify trade diversion effects. More studies have studied how to analyze panel data using the gravity models (Egger and Pfaffermayr, 2003; Serlenga and Shin, 2007). In addition, Xiong and Beghin (2012) adopted the gravity model to analyze how the MRL standards affect the bilateral food trade between the U.S. and Canada. They concluded that imports to the U.S. are negatively affected by the stringency of MRL standards. Following Drogue and DeMaria (2012), we specify the gravity model as follows: lnYijt = β0 + β1ln(gdpjt) + β2ln(popjt) + β3ln(gdpt) + β4ln(jpopt) + β5ln(ExchangeRatejt) + β6indexij +β7tariffit + β8ln(disj) + β9Asianj + εit (2) Yijt: the amount of imported vegetable i from country j at time t; gdpjt: exporter country j’s gross domestic product (GDP) at time t; popjt: exporter country j’s population at time t; jgdpt: Japan’s GDP at time t; jpopt: Japan’s population at time t; ExchangeRatejt: Exchange rate between exporter j and Japan at time t; indexij: the similarity index for vegetable i and country j; tariffit: Japan’s tariffs for vegetable i at time t; distj: geographic distance between exporter j and Japan; International Food and Agribusiness Management Review

162

Choi and Yue

Volume 20, Issue 1, 2017

Asianj: dummy variable indicating whether exporter j is in Asia; εit: an error term for vegetable i at time t, and it is normally distributed. Some papers also include the common border dummy variable to capture if the exporting and importing countries share common land borders (Drogue and DeMaria, 2010). Because Japan is an island country and does not share common borders with any country, this study includes the dummy variable ‘Asianj’ to capture if the exporter j is from the same continent as Japan. The expected signs of coefficients for the exporter j and Japan’s GDP and population are positive. That is, the higher the GDP and population of an exporting country, the more vegetables Japan imports from that country. Negative signs are anticipated for the exchange rate and distance. The signs for tariffs are expected to be negative and the signs for the similarity of MRL standards are expected to be positive. Vegetables used in the analysis and data description More than 40% of Japan’s imported vegetables are fresh (Dyck and Ito, 2004). Figure 1 shows that from 2008 to 2010, Japan’s leading imported vegetables by value were onions, peppers, cabbage, garlic, and carrots. During this three-year period, the value of the three main imported vegetables – onions, peppers, and cabbage – was more than $700 million in total. According to the Japanese Ministry of Agriculture, Forestry, and Fisheries (2012), cabbage, onions, radishes, Irish potatoes, and carrots had the highest consumption in the Japanese vegetable market. Figure 2 shows the quantity of Japanese domestic vegetable production and sales, which shows that Irish potatoes, cabbage, and onions were the top three most produced vegetables from 2001 to 2010. Irish potatoes, cucumbers, and eggplant were imported from only one or two countries, including South Korea and China, so they were excluded from this analysis. In addition, the quantities of imported peas and beans were too small to analyze the effects of MRL stringency on trade so they were also excluded from this analysis. 350,000 300,000

Value (×1000 USD)

250,000 200,000 150,000 100,000 50,000 0

onion

pepper cabbage

garlic

carrot

radish

tomato

bean

Import 313,832 283,742 194,069 82,359

81,812

68,554

24,323

13,168

pea

Irish eggplant lettuces potatoes 11,287 1,695 1,620 915

mushroom 793

cucumber 380

Figure 1. Japan’s total vegetable imports for the each of the top imported vegetables between 2008 and 2010 (adapted from Japanese Ministry of Agriculture, Forestry, and Fisheries (2013)). International Food and Agribusiness Management Review

163

Choi and Yue

Volume 20, Issue 1, 2017 3,000

Production Sale

Quantity (1000 ton)

2,500

2,000

1,500

1000

500

Iri s

ta ca toe bb s ag e on s io ns sw ee rad t p is ot h a to toes m cu ato cu es m be ca rs rro le ts ttu eg ce gp s la sp nts i pu nac m h pk pe ins pp er s be an s pe as

Vegetable crop

Figure 2. Japan’s average quantity of production and sale for the top produced vegetables (2001-2010) (adapted from Japanese Ministry of Agriculture, Forestry, and Fisheries, 2012). Tomatoes, peppers, lettuce, cabbage, onions, garlic, carrots, and radishes were included in this analysis because they are among the top 15 imported and domestically produced vegetables in Japan, and they come from the broadest array of other countries according to the Ministry of Agriculture, Forestry, and Fisheries (2013). The eight vegetables are classified into four types: (1) fruit vegetables – tomatoes and peppers; (2) leafy vegetables – lettuce and cabbage; (3) bulb vegetable – onions and garlic; and (4) root vegetables – carrots and radishes. We adopted a classification system based upon botany and the edible part method used by Yamaguchi (1983), who grouped vegetables by the part of the vegetable people eat. The classification of the four types of vegetables matches well with the pesticides used on each type. Peirce (1987) found that the major disease problems for peppers are the same as those for tomatoes. Thus, tomato and pepper production needs to use the same chemical components of pesticides. Moreover, Deshpande and Salunkhe (1998) pointed out that the most serious pests for lettuce are the same as those for cabbage, and as such the same chemical components of pesticide are used for lettuce and cabbage production. In addition, carrots and radishes are free from most diseases (Kotecha et al., 1998), but suffer from some root rot diseases (Masalkar and Keskar, 1998). Alliums including onions and garlic are susceptible to the same diseases and insects (Peirce, 1987). We analyzed how the similarities in MRL standards between Japan and other countries affected Japan’s vegetable imports between 1996 and 2010. The trade data were obtained from the United Nations Commodity Trade Statistics Database (Comtrade, 2013) of the United Nations Conference on Trade and Development (UNCTAD). The study used HS2 – 1996 6-digit codes for fresh tomatoes (070200), peppers (070960), lettuce (070519), cabbage (070490), onions (070310), garlic (070320), carrots (070610), and radishes (070690). Data on total vegetables (070000) were also collected from the same source.

International Food and Agribusiness Management Review

164

Choi and Yue

Volume 20, Issue 1, 2017

GDP and population data for exporters and Japan were retrieved from the World Bank Database (World Bank, 2013). In addition, exchange rate data were obtained from the Federal Reserve Bank (2013) and OANDA (2013). The distance between countries was calculated in miles and collected from the MapCrow database (MapCrow, 2013) based on the distance between the capitals. Each vegetable’s tariff and total tariffs were obtained from the World Trade Organization (2013d). According to these data, Japanese tariffs for vegetables have declined since 2000. Different countries have different MRL standards for vegetables. More than 30 countries’ MRL standards were obtained from the Homologa (2011) database including Argentina, Australia, Bolivia, Brazil, Cameroon, Canada, Chile, China, Columbia, Egypt, European Union, Ghana, India, Israel, Japan, Kenya, Malaysia, Mexico, Morocco, New Zealand, Norway, Philippines, Russia, Singapore, South Africa, South Korea, Switzerland, Tanzania, Thailand, Togo, Turkey, Ukraine, USA, Vietnam, and Zimbabwe.

3. Estimation results and discussion Table 1 describes the similarity indexes between Japanese vegetable MRL standards and MRL standards for the 34 countries. The Codex International Food Safety Standards are set by the Codex Alimentarius Commission. The Codex uses science-based reference levels and is provided to countries so that they can set their minimum levels of standards (Li and Beghin, 2012). Bolivia, Cameroon, Columbia, Egypt, Ghana, Kenya, Morocco, Tanzania, Togo, and Zimbabwe have adopted the Codex level of MRL standards for their vegetables. From Table 1, we can see Japan’s MRL standards are the least similar to the Codex, Mexico, Malaysia, and South Africa and they are the most similar to the standards of Chile, China, European Union, South Korea, Switzerland, Turkey, and Ukraine. The average of the indexes is about 0.53, which is much lower than one. This indicates Japan’s MRL standards are much stricter, by far, than those of other countries. Additionally, Japan’s MRL standards are quite different from the Codex standards, which means that Japan’s standards are far stricter than science-based reference levels. The panel data were analyzed using the Pooled Ordinary Least Square (POLS) model and Ordinary Least Square model with Random Effect (RE) in order to estimate how the similarity of MRL standards affects trade (Scheepers et al., 2007). Because some of the MRL standards and the distance have the same values throughout the years, this study did not adopt the fixed effect model and the first difference model. Additionally, serial correlations were detected for total vegetables, cabbage, onions, and carrots. Possible explanation for this is that εit in Equation 2 could be rewritten as εit = ui + vit. ui could affect explanatory variables positively or negatively. Thus, the RE model was adapted to adjust the correlations under main assumption that ui is not correlated to explanatory variables. Min and Choi (2009) demonstrated that the RE estimation is more consistent and effective when there is correlation and when the main assumption above holds. The RE estimation also has an advantage when it comes to estimating the coefficients of time invariant variables such as the similarity index and distance. First, total vegetables were analyzed in order to understand how similarities in MRL standards influence Japan’s overall vegetable imports. Table 2 shows the results of both POLS and RE estimations for 396 observations. The results indicate that the similarity index of MRL standards between Japan and other nations significantly affects the imports of total vegetables at a 1 or 5% significance level and in a positive way. Most importantly, in RE model, the similarity index affects total vegetable imports at 5% significance level. This means that if Japan’s MRL standards were not as strict as they are at present, vegetable imports would greatly increase. Generally, RE model has smaller t-statistics than POLS because RE adjusted the correlations. Tariff does not affect the vegetable imports significantly, but the exporters’ GDP and population significantly influence Japan’s vegetable imports. On the other hand, the exchange rate and Japan’s GDP and population do not have significant impacts on Japan’s vegetable imports. Asian dummy variable and distance are not statistically significant in the RE model. International Food and Agribusiness Management Review

165

Choi and Yue

Volume 20, Issue 1, 2017

Argentina Philippine, Singapore, Vietnam Austria Brazil Canada Codex Switzerland Chile China European Union India Israel South Korea Mexico Malaysia Norway New Zealand Russia Thailand Turkey Ukraine USA South Africa

0.55 0.54 0.55 0.55 0.55 0.47 0.59 0.57 0.60 0.58 0.48 0.51 0.57 0.48 0.49 0.50 0.55 0.51 0.54 0.56 0.56 0.54 0.45

0.53 0.55 0.51 0.58 0.54 0.44 0.55 0.59 0.58 0.57 0.47 0.56 0.59 0.48 0.51 0.51 0.54 0.51 0.57 0.58 0.51 0.54 0.47

0.57 0.54 0.55 0.52 0.54 0.48 0.54 0.52 0.54 0.57 0.50 0.50 0.57 0.50 0.50 0.50 0.53 0.50 0.48 0.52 0.50 0.52 0.47

0.59 0.56 0.54 0.56 0.54 0.46 0.59 0.53 0.56 0.59 0.53 0.51 0.60 0.52 0.50 0.51 0.53 0.57 0.50 0.55 0.54 0.54 0.48

0.53 0.50 0.52 0.57 0.52 0.47 0.55 0.56 0.56 0.57 0.47 0.49 0.58 0.50 0.51 0.50 0.53 0.50 0.53 0.53 0.56 0.52 0.43

0.54 0.60 0.48 0.52 0.55 0.50 0.56 0.59 0.55 0.59 0.46 0.47 0.53 0.47 0.48 0.51 0.48 0.58 0.54 0.53 0.55 0.53 0.48

0.54 0.58 0.49 0.51 0.56 0.48 0.55 0.58 0.56 0.58 0.47 0.50 0.54 0.48 0.49 0.51 0.49 0.52 0.53 0.55 0.56 0.54 0.45

Table 2. Gravity model estimation results for Japanâ&#x20AC;&#x2122;s total vegetable imports.1,2,3 Total vegetable imports

ln_gdp ln_pop ln_jgdp ln_jpop ln_ExchangeRate total tariff total index ln_dist Asian Constant Number of observations

pooled OLS

0.0801*** (2.73) 0.598*** (5.11) -0.971 (-0.63) 7.141 (0.17) 0.0843** (2.18) 0.114 (0.25) 25.70*** (7.84) -1.613*** (-7.67) -1.251*** (-3.21) -102.9 (-0.13) 396

0.171 (1.42) 0.891**(2.25) -0.605 (-1.11) -12.30 (-0.77) 0.0324 (0.75) 0.0722 (0.56) 29.09** (2.51) -0.983 (-1.20) -0.597 (-0.41) 234.1 (0.76) 396

= P<0.10; ** = P< 0.05; *** = P<0.01. t-statistics in parentheses. 3 OLS = Ordinary least square; RE = Random effect. 2

International Food and Agribusiness Management Review

166

0.58 0.55 0.54 0.53 0.51 0.46 0.58 0.59 0.55 0.56 0.49 0.53 0.52 0.49 0.50 0.50 0.53 0.53 0.53 0.57 0.52 0.53 0.43

Radish

Carrot

Garlic

Onion

Cabbage

Lettuce

Pepper

Tomato

Country

Total

Table 1. The similarity index for vegetables between Japan and other countries.

0.57 0.55 0.54 0.53 0.51 0.47 0.57 0.57 0.55 0.55 0.46 0.51 0.52 0.47 0.48 0.51 0.54 0.51 0.53 0.55 0.56 0.52 0.43

Choi and Yue

Volume 20, Issue 1, 2017

Correlations were not identified for the fruit vegetables: tomatoes and peppers. Thus, the results of the POLS and RE estimations are the same. Although coefficients are the same, t-statistics are smaller in RE estimations which provide more accurate standard errors. As shown in Table 3, the similarity index significantly and positively impacts Japan’s imports for tomatoes and peppers at a 1 and 5% significance level, respectively. The indexes have greater influence on Japanese tomato and pepper imports than on total vegetable imports because the coefficients of tomatoes (61.89) and peppers (38.78) for the similarity indexes are much larger than that of total vegetables (29.09). Hence, for fruit vegetables, the strict MRL standards are influential when it comes to controlling the quantities of imports. Exporter’s GDPs are significant in both estimation results, but the signs of the coefficient are opposite. As expected, the signs of GDP for tomatoes are positive. However, the signs of GDP for pepper are negative. As expected, distance affects the tomato and pepper imports significantly and negatively, which means if distances between Japan and exporters are large, the fruit vegetable imports decreases. The exchange rates and Japan’s population are not important factors affecting vegetable imports. The Asian dummy variable negatively affects Japan’s tomato and pepper imports, which means Japan imported more tomatoes and peppers from non-Asian countries. Table 4 presents the leafy vegetables’ estimation results. The coefficients of the similarity indexes are significant at a 1% significance level and they are positive for both lettuce and cabbage imports as expected. In other words, if Japan’s MRL standards were more similar to those of the exporters, the quantities of lettuce and cabbage imports would increase. For leafy vegetables, the exporters’ GDP and population, Japan’s GDP, and the Asian dummy variable significantly affect imports while other variables such as Japan’s population, distance, and exchange rate are not significant. The estimation results for bulb vegetables and root vegetables are summarized in Table 5 and Table 6. The similarity indexes for total vegetables, fruit, and leafy vegetables indicate that the indexes affect the imports significantly and positively. However, the coefficients of the similarity indexes for bulb vegetables and root vegetables are not significant, so the impact of the indexes on imports of bulb vegetables and root vegetables is negligible. Table 3. Gravity model estimation results for Japan’s tomato and pepper imports.1,2,3 Tomato pooled OLS

Pepper RE

pooled OLS

1.072*** (7.61) ln_gdp 1.072*** (4.89) ln_pop 2.199*** (6.78) 2.199*** (9.59) ln_jgdp 1.767(0.87) 1.767 (0.57) ln_jpop -139.8* (-1.85) -139.8 (-1.61) ln_ExchangeRate 0.114 (0.53) 0.114 (0.49) ** tomato tariff -2.535 (-2.31) -2.535** (-2.49) ** tomato index 61.89 (2.46) 61.89*** (5.11) *** ln_dist -57.19 (-6.98) -57.19*** (-7.32) *** Asian -122.6 (-7.26) -122.6*** (-7.75) pepper tariff pepper index Constant 2,967.0** (2.07) 2,967.0* (1.85) Number of observations 71 71

-0.526*** (-20.66) -0.177 (-1.11) 1.853* (1.81) 54.52 (1.16) -0.394 (-1.43)

-2.803*** (-3.46) -2.272* (-1.89) -0.346 (-0.52) 38.78*** (5.35) -1,038.4 (-1.19) 67

= P<0.10; ** = P< 0.05; *** = P<0.01. t-statistics in parentheses. 3 OLS = Ordinary least square; RE = Random effect. 2

International Food and Agribusiness Management Review

167

RE -0.526*** (-31.87) -0.177 (-1.04) 1.853** (1.97) 54.52 (0.71) -0.394 (-1.33)

-2.803*** (-3.31) -2.272* (-1.86) -0.346 (-0.51) 38.78*** (4.85) -1,038.4 (-0.71) 67

Choi and Yue

Volume 20, Issue 1, 2017

Table 4. Gravity model estimation results for Japan’s lettuce and cabbage imports.1,2,3 Lettuce pooled OLS

Cabbage RE

pooled OLS

0.604*** (14.64) ln_gdp 0.604*** (8.26) *** ln_pop 0.564 (2.86) 0.564*** (5.31) ln_jgdp -6.349** (-2.40) -6.349** (-2.08) ln_jpop 38.13 (0.36) 38.13 (0.23) ln_ExchangeRate -0.0531 (-0.24) -0.0531 (-0.32) lettuce tariff 0.837 (0.70) 0.837 (0.55) *** lettuce index 66.72 (4.45) 66.72*** (4.80) ln_dist -0.0354 (-0.05) -0.0354 (-0.06) Asian -5.317*** (-4.69) -5.317*** (-3.62) cabbage tariff cabbage index Constant -579.2 (-0.30) -579.2 (-0.18) Number of observations 41 41

0.544*** (24.07) 0.879*** (8.01) -4.990*** (-3.29) -84.86 (-1.36) -0.136** (-2.02)

-0.591 (-1.56) -3.176*** (-4.75) -0.477 (-0.57) 34.14*** (7.06) 1,700.9 (1.46) 126

RE 0.459*** (7.46) 0.919*** (4.68) -4.346** (-2.57) -102.1* (-1.90) -0.212* (-1.86)

-0.439 (-0.52) -2.638*** (-3.04) -0.742 (-1.17) 33.51** (2.42) 2,005.2** (1.99) 126

= P<0.10; ** = P< 0.05; *** = P<0.01. t-statistics in parentheses. 3 OLS = Ordinary least square; RE = Random effect. 2

Table 5. Gravity model estimation results for Japan’s onion and garlic imports.1,2,3 Onion pooled OLS

Garlic RE

pooled OLS

0.128 (0.36) ln_gdp 0.223*** (4.10) ln_pop -0.0565 (-0.28) -0.00255 (-0.01) ln_jgdp 0.767 (0.38) 0.812 (1.49) ln_jpop 23.44 (0.62) -24.18 (-0.86) ln_ExchangeRate 0.759*** (6.96) 0.757** (1.98) onion tariff 0.0421 (0.34) 0.0884 (1.35) onion index 11.64 (1.14) 14.04 (0.69) ln_dist -0.00396 (-0.01) -0.435 (-0.37) Asian 2.186*** (3.62) 3.420 (1.28) garlic tariff garlic index Constant -459.6 (-0.64) 429.4 (0.82) Number of observations 153 153

0.266*** (3.88) 0.280 (1.47) 7.217*** (4.60) -105.9 (-1.33) 0.940*** (6.96)

-0.864 (-0.48) 4.041 (1.07) -0.199 (-0.22) 17.50 (1.05) 1,759.8 (1.18) 61

RE 0.266*** (3.28) 0.280 (1.46) 7.217** (2.39) -105.9 (-1.31) 0.940*** (6.38)

-0.864 (-0.42) 4.041 (0.94) -0.199 (-0.26) 17.50 (0.95) 1,759.8 (1.18) 61

= P<0.10; ** = P< 0.05; *** = P<0.01. 2 t-statistics in parentheses. 3 OLS = Ordinary least square; RE = Random effect.

In both cases, the similarity indexes are affected positively but insignificantly. We can see that an exporter’s GDP and exchange rate are the most significant among the other variables for bulb vegetables in Table 5. Japanese population, tariff, distance and even the similarity index do not have significant impacts on bulb vegetable imports. The similar results are shown for root vegetables in Table 6. Root vegetables’ similarity index does not play any important role to affect root vegetable imports. Exporter’s GDP, exchange rate, distance, and tariff for radish are somewhat significant for root vegetable imports. International Food and Agribusiness Management Review

168

Choi and Yue

Volume 20, Issue 1, 2017

Table 6. Gravity model estimation results for Japan’s carrot and radish imports.1,2,3 Carrot pooled OLS

Radish RE

pooled OLS

0.361*** (2.81) ln_gdp 0.342*** (4.67) ** ln_pop -0.296 (-2.42) 0.194 (0.38) ln_jgdp 1.377 (0.69) -1.207 (-0.60) ln_jpop 18.36 (0.19) -10.45 (-0.15) ln_ExchangeRate 0.458*** (3.08) 0.218 (0.77) carrot tariff 0.280 (0.21) -0.0888 (-0.09) carrot index 19.95 (0.68) 3.039 (0.08) ln_dist -2.416*** (-3.21) -2.038* (-1.71) Asian -2.344 (-1.26) -3.396* (-1.69) radish tariff radish index Constant -365.4 (-0.20) 245.7 (0.18) Number of observations 84 84

0.0264 (0.48) 0.389** (2.40) -1.714 (-0.91) -85.63 (-0.84) 0.490*** (4.27)

0.0264 (0.35) 0.389 (1.50) -1.714 (-0.89) -85.63 (-0.72) 0.490*** (3.44)

-2.401*** (-3.66) 0.152 (0.12) -2.848** (-2.59) 42.36 (1.17) 1,654.8 (0.86) 68

-2.401*** (-3.08) 0.152 (0.11) -2.848** (-1.98) 42.36 (1.06) 1,654.8 (0.73) 68

= P<0.10; ** = P< 0.05; *** = P<0.01. t-statistics in parentheses. 3 OLS = Ordinary least square; RE = Random effect. 2

Overall, the RE methods gives smaller t-statistics than Pooled OLS in most cases since the RE adjusted the correlations. Although correlation has been detected for some of the vegetables, we found the results of the RE estimation are very close to those of POLS. Also, for all vegetables in this study, tariffs are not as effective as the MRL standards to control vegetable imports in Japan. Table 7 represents all 8 vegetable RE results to compare each other. In sum, the results show that Japan’s strict MRL standards significantly impacted its vegetable imports, especially fruit and leafy vegetable imports. The higher the similarity index, the more similar Japan’s MRL standards are to other countries, and the less strict the Japanese MRL standards are compared to other countries. Thus, Japan’s stricter MRL standards significantly reduced imports for fruit and leafy vegetables, which is consistent with findings from previous research (Burnquist et al., 2013; Drogue and DeMaria, 2012, 2013). However, Japan’s MRL standards did not significantly reduce the imports of bulb and root vegetables. An interesting result is that the MRL standards’ impacts on vegetable trades differ across different types of vegetables. Many studies have shown that strict MRL standards could have greater impacts on imports than tariffs, quotas, and subsidies (Bao and Qiu, 2009; Drogue and DeMaria, 2012; Hejazi et al., 2016). Hence, foreign vegetable producers should consider MRL standards in their business decisions. Our findings suggest producers in countries with higher similarity in MRL standards to Japan can increase fruit and leafy vegetable exports to Japan. Both foreign and domestic vegetable producers striving to meet Japan’s high MRL standards have to bear MRL compliance costs, which lead to higher production costs and thus decrease the profit margins. However, for countries where the input costs are lower than Japan (for example, Vietnam and Philippine have lower labor costs; Australia and Canada have lower land costs), they can comply with the MRL standards with much lower costs. The lower compliance costs can help keep vegetable price lower than the vegetables produced in Japan. Our findings also indicate the MRL standards affect the imports of different kinds of vegetables differently. By improving the MRL levels and make them more similar to Japan while maintaining the compliance costs low, foreign fruit and leafy vegetable producers but not root vegetable products can potentially increase their vegetable exports.

International Food and Agribusiness Management Review

169

Choi and Yue

Volume 20, Issue 1, 2017

Table 7. Gravity model estimation RE results for 8 vegetable imports.1,2 Fruit tomato 1.072*** (7.61) ln_pop 2.199*** (9.59) ln_jgdp 1.767 (0.57) ln_jpop -139.8 (-1.61) ln_ExchangeRate 0.114 (0.49) tariff -2.535** (-2.49) MRL index 61.89*** (5.11) ln_dist -57.19*** (-7.32) Asian -122.6*** (-7.75) Constant 2,967.0* (1.85) Number of 71 observations

ln_gdp

Leafy

Bulb

Root

pepper

lettuce

cabbage

onion

garlic

carrot

radish

-0.526*** (-31.87) -0.177 (-1.04) 1.853** (1.97) 54.52 (0.71) -0.394 (-1.33) -0.346 (-0.51) 38.78*** (4.85) -2.803*** (-3.31) -2.272* (-1.86) -1,038.4 (-0.71) 67

0.604*** (14.64) 0.564*** (5.31) -6.349** (-2.08) 38.13 (0.23) -0.0531 (-0.32) 0.837 (0.55) 66.72*** (4.80) -0.0354 (-0.06) -5.317*** (-3.62) -579.2 (-0.18) 41

0.459*** (7.46) 0.919*** (4.68) -4.346** (-2.57) -102.1* (-1.90) -0.212* (-1.86) -0.742 (-1.17) 33.51** (2.42) -0.439 (-0.52) -2.638*** (-3.04) 2,005.2** (1.99) 126

0.128 (0.36) -0.00255 (-0.01) 0.812 (1.49) -24.18 (-0.86) 0.757** (1.98) 0.0884 (1.35) 14.04 (0.69) -0.435 (-0.37) 3.420 (1.28) 429.4 (0.82) 153

0.266*** (3.28) 0.280 (1.46) 7.217** (2.39) -105.9 (-1.31) 0.940*** (6.38) -0.199 (-0.26) 17.50 (0.95) -0.864 (-0.42) 4.041 (0.94) 1,759.8 (1.18) 61

0.361*** (2.81) 0.194 (0.38) -1.207 (-0.60) -10.45 (-0.15) 0.218 (0.77) -0.0888 (-0.09) 3.039 (0.08) -2.038* (-1.71) -3.396* (-1.69) 245.7 (0.18) 84

0.0264 (0.35) 0.389 (1.50) -1.714 (-0.89) -85.63 (-0.72) 0.490*** (3.44) -2.848** (-1.98) 42.36 (1.06) -2.401*** (-3.08) 0.152 (0.11) 1,654.8 (0.73) 68

1* 2

= P<0.10; ** = P< 0.05; *** = P<0.01. t-statistics in parentheses.

4. Conclusions Food safety has presented a vexing problem in many countries, and as such, countries have attempted to control food-borne illness by improving the food safety quality of both domestic production and imports. In this regard, MRL standards have played a significant role. The extent to which MRL standards affect the trade of agricultural products has been widely studied. Few studies have examined how MRL standards affect the trade of different types of food differently. This study is intended to fill this knowledge gap in the literature by exploring the impact of MRL standards on the imports of different types of vegetable in Japan. Using the widely used gravity model, this study analyzed how the MRL standards affect vegetable imports in Japan. The similarity index was adopted to evaluate the degree to which Japan’s MRL standards are different or similar to other countries’ standards. In addition, tariff, GDP, population, distance, and the exchange rate were included as explanatory variables. POLS and RE estimations were compared and applied to estimate the gravity model. The main finding of this research is that total vegetables, fruit vegetables, and leafy vegetables are significantly affected by the degree of similarity in MRL standards between traders. By contrast, the similarity of MRL standards does not significantly affect Japan’s imports of bulb vegetables and root vegetables such as onions and carrots. This study finds that the type of vegetables makes a difference when it comes to how the MRL standards affect imports. Thus, by setting strict MRL standards, Japan could effectively restrict fruit and leafy vegetable imports. From the exporters’ perspective, the exporters have suffered great loss by the significantly International Food and Agribusiness Management Review

170

Choi and Yue

Volume 20, Issue 1, 2017

reduced vegetable exports to Japan. In this study, we have not considered the exporters’ actual costs as a result of complying with Japan’s MRL standards. If these costs were considered, the exporters’ loss would be even greater. Therefore, we concluded that the stringency of MRL standards has impeded Japanese imports for fruit and leafy vegetables (but not for bulb and root vegetables). If the Japan’s MRL standards were less strict, fruit and leafy vegetable imports would greatly increase. On the other hand, exporters to Japan can set stricter MRL standards on fruit and leafy vegetables to increase their exports to Japan. This study contributes to the empirical literature in understanding how Japan’s food standards affect the imports of the four types of vegetables: fruit, leafy, bulb and root vegetables. If the cost of implementing and adjusting the MRL standards could be measured, it would help further explain how the MRL standards would affect producer and consumer prices, and thus the profits of exporters. This could be an interesting future research topic. How the stringency of MRL standards affects the consumption volume of fruits and vegetables, and thus human health could be another interesting future research topic.

References Achterbosch, T.J., A. Engler, M.L. Rau, and R. Toledo. 2009. Measure the measure: the impact of differences in pesticide MRLs on Chilean fruit exports to the EU. In: International Association of Agricultural Economists Conference, Beijing, China, pp. 16-22. Available at: http://tinyurl.com/gpf5fsv. Anderson, K. 2009. Terroir rising? Varietal and quality distinctiveness of Australia? Wine Regions (No. 2009-18). University of Adelaide, School of Economics. Available at: http://tinyurl.com/hxbzevc. Anderson, K. 2010. Varietal intensities and similarities of the world’s wine regions (No. 2010-04). University of Adelaide, Wine Economics Research Centre. Available at: http://tinyurl.com/z7m8jtw. Bao, X. and L.D. Qiu. 2009. Quantifying the trade effects of technical barriers to trade: evidence from China. In: Asia Pacific Trade Seminars, Hong Kong (A1620). Available at: http://tinyurl.com/hvjjuvk. Burnquist, H.L., K. Shutes, M. Rau, M.J. Pinto de Souza, and R. Nunes de Faria. 2011. Heterogeneity index of trade and actual heterogeneity index – the case of maximum residue levels for pesticides. In: Agricultural and Applied Economics Association’s 2011 Annual Meeting, Pittsburgh, Pennsylvania, USA. Available at: http://tinyurl.com/jsucz4o. Carrère, C. 2006. Revisiting the effects of regional trade agreements on trade flows with proper specification of the gravity model. European Economic Review 50: 223-247. Comtrade. 2013. Japanese vegetable imports. Available at: http://comtrade.un.org/db/. Deshpande, S.S. and D.K. Salunkhe. 1998. Lettuce. In: Handbook of vegetable science and technology, edited by D.K. Salunkhe and S.S. Kadam. Marcel Dekker Inc., New York, NY, USA. Drogue, S. and F. DeMaria. 2010. Pesticides residues and trade: the apple of discord? Food Policy 37: 641-649. Drogue, S. and F. DeMaria. 2012. Comparing apples with pears. How differences in pesticide residues regulations impact trade? (No. 201201). UMR MOISA: Marchés, Organisations, Institutions et Stratégies d’Acteurs: CIHEAM-IAMM, CIRAD, INRA, Montpellier SupAgro, IRD-Montpellier, France. Available at: http://tinyurl.com/jehq5f7. Drogue, S. and F. DeMaria. 2013. Apples, Pears and pesticides impact of heterogeneous regulations governing pesticide residues on world trade. Inra Sciences Sociales 2013: 3. Dyck, J.H. and K. Ito. 2004. Japan’s fruit and vegetable market. In: Wu Huang, S. (ed.) Global Trade Patterns in Fruits and Vegetables. USDA-ERS Agriculture and Trade Report WRS-04-06. USDAERS, Washington, DC, USA, pp. 64-76. Egger, P. and M. Pfaffermayr. 2003. The proper panel econometric specification of the gravity equation: a three-way model with bilateral interaction effects. Empirical Economics 28: 571-580. Federal Reserve Bank. 2013. Exchange rates. Available at: http://www.federalreserve.gov/releases. Food and Agriculture Organization and World Health Organization. 2006. Food safety risk analysis – a guide for national food safety authorities. FAO Food and Nutrition, paper 87. Available at: http:// tinyurl.com/zcwozeh.

International Food and Agribusiness Management Review

171

Choi and Yue

Volume 20, Issue 1, 2017

Hejazi, M., J.H. Grant and E. Peterson. 2016. Hidden trade costs? Maximum residue limits and US exports to trans-Atlantic and trans-Pacific trading partners. In: Agricultural and Applied Economics Association Annual Meeting, Boston, MA July 31 – August. Available at: http://tinyurl.com/gvq86rl. Homologa. 2011. MRL standards. Available at: http://www.homologa.com. Jaffe, A. (1986). Technological opportunity and spillovers of R&D: evidence from firms’ patents, profits, and market value. American Economic Review 76: 984-1001. Japanese Ministry of Agriculture, Forestry, and Fisheries. 2012. Major Vegetables. Available at: http:// tinyurl.com/jgq3wzl. Japanese Ministry of Agriculture, Forestry, and Fisheries. 2013. Foreign Trade of Agricultural, Forestry and Fishery Products. Available at: http://tinyurl.com/jgq3wzl. Kotecha, P.M., B.B. Desai, and D.L. Madhavi. 1998. Carrots. In: Handbook of vegetable science and technology, edited by D.K. Salunkhe and S.S. Kadam. Marcel Dekker Inc., New York, NY, USA, pp. 119-140. Li, Y. and J. Beghin. 2012. Protectionism indices for non-tariff measures: an application to maximum residue levels. Report No. 34985. Iowa State University, Ames, IA, USA. Liu. L. and C. Yue. 2012. Investigating the impact of MRL standards’ similarity on trade. In: Non-tariff measures with market imperfections: trade and welfare implications, edited by J.C. Beghin. Emerald Press, Castle Hill, Australia, pp.151-164. MapCrow. 2013. Distances. Available at: http://www.mapcrow.info/. Martinez, L.R. and S. Thornsbury. 2010. Identifying maximum residue limit (MRL) regulations faced by Michigan fruit industries. (No. 136242). Michigan State University, Department of Agricultural, Food, and Resource Economics. Michigan, MI, USA. Masalkar, S.D. and B.G. Keskar. 1998. Other roots, tubers, and rhizomes. In: Handbook of vegetable science and technology, edited by D.K. Salunkhe and S.S. Kadam. Marcel Dekker Inc., New York, NY, USA. Min, I. and P. Choi. 2009. STATA – analyzing panel data. The Korean Association of STATA. Seoul, South Korea. Oanda. 2013. Exchange rates. Available at: http://www.oanda.com/currency/. Otsuki, T., J. Wilson and M. Sewadeh. 2001a. What price precaution? European harmonisation of aflatoxin regulations and African groundnuts exports. European Review of Agricultural Economics 28: 263-284. Otsuki, T., J. Wilson and M. Sewadeh. 2001b. Saving two in a billion: quantifying the trade effect of European food safety standards on African exports. Food Policy 26: 495-514. Peirce, L.C. 1987. Vegetables: characteristics, production, and marketing. John Wiley and Sons, New York, NY, USA. Scheepers, S., A. Jooste and Z.G. Alemu. 2007. Quantifying the impact of phytosanitary standards with specific reference to MRLs on the trade flow of South African avocados to the EU. Agrekon 46: 260-273. Serlenga, L. and Y. Shin. 2007. Gravity models of intra-EU trade: application of the CCEP-HT estimation in heterogeneous panels with unobserved common time-specific factors. Journal of Applied Econometrics. 22: 361-381. Silva, J.S. and S. Tenreyro. 2006. The log of gravity. Review of Economics and Statistics 88: 641-658. Tinbergen, J. 1962. Shaping the world economy: suggestions for an international economic policy. Twentieth Century Fund, New York, NY, USA. Wei, G., J. Huang and J. Yang. 2012. Honey safety standards and its impacts on china’s honey export. Journal of Integrative Agriculture 11: 684-693. Wilson, J.S. and T. Otsuki. 2004. To spray or not to spray: pesticides, banana exports, and food safety. Food Policy 29: 131-146. Winchester, N., M-L. Rau, C. Goetz, B. Larue, T. Otsuki, K. Shutes, C. Wieck, H.L. Burnquist, M. Souaza and R. Faria. 2012. The impact of regulatory heterogeneity on agri-food trade. The World Economy 35: 973-993. World Bank. 2013. GDP and population. Available at: http://data.worldbank.org/indicator/. World Trade Organization (WTO). 2013a. Japan – measures affecting agricultural products, dispute settlement: DISPUTE DS76. Available at: http://tinyurl.com/jqupef6.

International Food and Agribusiness Management Review

172

Choi and Yue

Volume 20, Issue 1, 2017

World Trade Organization (WTO). 2013b. Japan – measures affecting imports of pork, dispute settlement: DISPUTE DS66. Available at: http://tinyurl.com/h8ztvn9. World Trade Organization (WTO). 2013c. Japan – measure affecting the importation of apples, dispute settlement: DISPUTE DS245. Available at: http://tinyurl.com/gw37rc6. World Trade Organization (WTO). 2013d. Tariffs. Available at: http://tinyurl.com/j7s3uoo. Xiong, B. and J.C. Beghin. 2012. Stringent maximum residue limits, protectionism, and competitiveness: the cases of the U.S. and Canada. (No. 35584). Iowa State University Ames, IA, USA. Yamaguchi, M. 1983. World vegetables: principles, production and nutritive values. The AVI Publishing Company Inc., Westport, CT, USA. Yue, N., H. Kuang, L. Sun, L. Wu and C. Xu. 2010. An empirical analysis of the impacts of EU’s new food safety standards on China’s tea export. International Journal of Food Science and Technology 45: 745-750.

International Food and Agribusiness Management Review

173